ABSTRACT Title of Thesis: THERMAL CYCLING DESIGN ...

ABSTRACT

Title of Thesis: THERMAL CYCLING DESIGN ALTERNATIVES FOR THE

POLYMERASE CHAIN REACTION

Degree Candidate: Monte Lewis

Degree and Year: Master of Science, 2005

Thesis directed by: Associate Professor Keith E. Herold Department of Mechanical Engineering

PCR (polymerase chain reaction) is a process by which a small amount of DNA is

amplified many times to yield an easily detectable amount. This process is widely used

for the detection of bacterial pathogens for biodefense, in basic research, criminal

identification, and disease detection in humans. The reaction mixture must be cycled

repeatedly between three different temperature levels in PCR. The reaction mixture is

first heated at 94 °C, cooled to 54 °C, and then heated to 72 °C. This cycle is repeated

20-40 times.

The main objective of the work reported here is to evaluate alternative

heating/cooling schemes for PCR with the ultimate goal of speeding up a PCR reaction.

A secondary goal is to arrive at a design that is consistent with battery operation to allow

for a portable PCR device. Insight is gained about interactions between the PCR reaction

and the engineering system.

The first device analyzed for use as a thermal cycler is a Peltier thermoelectric

cooler. A single thermoelectric cooler is characterized. The maximum and average

heating rates and power consumption are determined through experiment. The

thermoelectric cooler is evaluated under different conditions and modeled with the

lumped capacitance method to reveal the characteristics of the device as a thermal cycler.

This Peltier device is used as a heating/cooling system for subsequent work on alternative

reaction vessels. A commercially available Peltier thermocycler is also analyzed to

determine the heating and cooling rates and power consumption. PCR results are

presented that show the time limits of the PCR reaction.

The second device analyzed for use as a thermal cycler is a convective

thermocycler. An advantage of the convective thermocycler is it allows for a wide range

of vessel geometries. Two variants of convective thermocyclers were built and tested for

performance. Heat transfer to a sample is modeled using finite element methods as well

as with the lumped capacitance method and corroborated with experiment. Reaction

vessels of different sizes and material are modeled in order to better understand the

relationship between the air velocity and heating characteristics of the thermal cycler and

the reaction volume.

The third device analyzed is a microheater system. The microheater design is

arrived at as the preferred design for high speed PCR. Microheaters were fabricated

using standard MEMS techniques. Problems encountered in the design, fabrication, and

use of these heaters are presented. The microheaters are evaluated for thermal cycling

performance under different conditions. Chambers fabricated out of polycarbonate are

tested for use as reaction volumes. Two dimensional finite element analysis is used to

model these systems and the models are corroborated with experiment. The behavior of

liquid in polycarbonate chambers undergoing thermal cycling is investigated. Problems

associated with thermal cycling in polycarbonate chambers is discussed along with

possible solutions.

THERMAL CYCLING DESIGN ALTERNATIVES FOR THE POLYMERASE CHAIN

REACTION

by

Monte Lewis

Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial fulfillment

of the requirements for the degree of Master of Science

2005 Advisory Committee: Associate Professor Keith E. Herold, Chair Associate Professor Elisabeth Smela Assistant Professor Bao Yang

ii

ACKNOWLEDGMENTS

I would like to thank my professor, Dr. Keith Herold, for giving me the opportunity to work for him, for his patience, and for his help in completing this project. I am very thankful to Dr. Jungho Kim for his help on the control system for the microheaters. The USDA funded this work, for which I am grateful. I was very lucky to have worked with our post doc from Ukraine, Andrey Matviyenko who spent many days doing PCR and working on mechanisms for PCR. Nikolai Sergeev and Avi Rasooly also provided valuable input to the project that I am thankful for. I would like to say thank you to Tom Laughran who taught me many things, was always very helpful and responsive, and set an excellent example for cleanroom behavior. Tom also graciously deposited metal on our samples. Nolan Balew and Jon Hummel also provided valuable support in the clean room. I would like to thank the many people I have met in the clean room and provided me with information including Steve Fanning, Samuel Moseley, and Remi Delille. Sayeed Moghaddam and Vytenis Benetis went out of their way to help me in the cleanroom and with the microheater design. I am also thankful to Dr. Elisabeth Smela who allowed the use of her lab equipment and was also an excellent teacher. I would like to thank Lou Hromalo for helping me in several situations. Corey Skurdal also provided valuable input and performed PCR many times. The Terrapin Masters swim team provided me with much needed exercise and camaraderie during my time here for which I am thankful.

iii

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... vi

LIST OF FIGURES ........................................................................................................ vii

SYMBOLS...................................................................................................................... xiii

1. INTRODUCTION TO PCR AND THERMOCYCLERS.....................................1

1.2 PCR reaction and detection reagents ........................................................5

1.3 Thermocycler metrics ...............................................................................7

2. TESTING OF PELTIER THERMOCYCLER FOR PCR.................................10

2.1 Testing of an independent TEC, apparatus and procedures....................10

2.2 Independent TEC testing results .............................................................13

2.2.1 Heating flux characteristics of TEC........................................................18

2.2.2 Cooling flux characteristics of TEC .......................................................26

2.2.3 Power consumption.................................................................................29

2.3 Independent TEC testing discussion.......................................................29

2.4 Testing of a commercially available Peltier thermocycler,

apparatus and procedures........................................................................31

2.4.1 Commercial Peltier thermocycler testing results ....................................33

2.4.2 Power consumption of the commercial device ......................................34

2.4.3 PCR results with the commercial device ................................................35

3. ANALYSIS OF CONVECTIVE THERMOCYCLERS FOR PCR...................42

3.1 Testing of a convective thermocycler, apparatus and procedures ..........42

3.2 Convective thermocycler test results ......................................................45

3.2.1 Power consumption of convective thermocyclers ..................................49

iv

3.2.2 Convection characteristics of the thermocycler......................................50

3.2.3 System response with a test sample ........................................................55

3.3 Convective thermocycler options to enhance performance ....................66

4. ANALYSIS OF MICROHEATERS FOR PCR...................................................77

4.1 Synopsis of work done on microheaters by other groups .......................77

4.1.1 Comparison of designs............................................................................82

4.2 Microheater design .................................................................................85

4.2.1 Electronic control system design ............................................................95

4.3 Microheater fabrication...........................................................................99

4.3.1 Microheater fabrication equipment and techniques................................99

4.3.2 Microheater fabrication sequence .........................................................101

4.3.3 Microheater fabrication results .............................................................103

4.4 Microheater heat treatment and calibration ..........................................106

4.5 Microheater testing ...............................................................................110

4.5.1 Microheater testing results....................................................................112

5. MICROCHAMBERS FOR PCR ........................................................................116

5.1 Microchamber design considerations ...................................................124

5.2 Microchamber fabrication equipment and techniques..........................134

5.3 Microchamber testing ...........................................................................136

5.3.1 Countersunk polycarbonate chambers ..................................................137

5.3.2 Glass capillaries ....................................................................................139

5.3.3 Thermally bonded polycarbonate chambers .........................................140

5.3.3.1 Thermally bonded polycarbonate chamber sealing ..............................145

v

6. SUMMARY, CONCLUSIONS AND FUTURE WORK.......................................149

6.1 Summary ...............................................................................................149

6.1.1 Testing of Peltier thermocyclers ...........................................................150

6.1.2 Testing of convective thermocyclers ....................................................150

6.1.3 Testing of a microheater thermocycler .................................................151

6.2 Conclusions...........................................................................................152

6.3 Future work...........................................................................................157

REFERENCES ...............................................................................................................159

vi

LIST OF TABLES

Table 1.1 Primer sequences used in amplification of B. anthracis ....................................6 Table 2.1 Properties of aluminum oxide used in the analysis..........................................18 Table 2.2 Comparison of TEC with and without cooling fan ..........................................29 Table 2.4 Protocol for elongation time experiment..........................................................35 Table 3.1 Results of air velocity analysis for single and double fan convective

thermocyclers...................................................................................................53 Table 3.2 Nusselt number and convection coefficients for single and double fan

convective thermocycler ..................................................................................55 Table 3.3 Calculated time constants (τ) for the air velocities within the single and double fan convective thermocycler for a modeled 200 µL polypropylene tube with 10 µL of water .........................................................57 Table 3.4 Polypropylene properties .................................................................................58 Table 3.5 Thermal conductivities of candidate materials for PCR reaction vessels ..............................................................................................................73 Table 3.6 Biot number for varying polypropylene cylindrical diameters ........................74 Table 4.1 Comparison of thermocycling devices.............................................................84 Table 4.2 Result of heat treatment on heater and temperature probe.............................107 Table 4.3 Comparison of microheater operation with and without heat sink ................115 Table 5.1 Model to predict pressure in sealed two-phase system ..................................131 Table 6.1 Comparison of thermocyclers ........................................................................149

vii

LIST OF FIGURES

Figure 1.1 One cycle of the PCR process.........................................................................2 Figure 1.2 A PCR cycle showing plateaus at 54, 72 and 94 °C .......................................2 Figure 1.3 Schematic of PCR cycle with dashed areas representing times included in the average power over cycle calculation ....................................8 Figure 2.1 Thermoelectric thermocycler setup...............................................................10 Figure 2.2 H-bridge configuration used in the control of the TEC ................................11 Figure 2.3 Side view of TEC setup ................................................................................12 Figure 2.4 Simulated PCR cycle for TEC without cooling fan , 13.8 V input...............14 Figure 2.5 Simulated PCR cycle for TEC with cooling fan on, 13.8V input.................14 Figure 2.6 Steady state heating mode of the TEC without cooling fan and 10.0 V input ..................................................................................................15 Figure 2.7 Steady state heating mode of the TEC with cooling fan and 10.0 V input ..................................................................................................16 Figure 2.8 Steady state cooling mode of the TEC without cooling fan and 13.8 V input............................................................................................16 Figure 2.9 Steady state cooling mode of the TEC with cooling fan and 13.8 V input ..................................................................................................17 Figure 2.10 Heat flux produced by thermoelectric cooler during heating from 72 to 94 °C and corresponding top and bottom of TEC temperature with 13.8 V input, cooling fan ..................................................20 Figure 2.11 Calculated Performance of TEC using hot and cold side temperatures from experiment ......................................................................21 Figure 2.12 2-D geometry used for cooling and heating model.......................................23 Figure 2.13 Calculated temperature for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan on.........................................................................................23 Figure 2.14 Heat flux produced by thermoelectric cooler during heating from 72 to 94 °C with 13.8 V input, cooling fan off ....................................24

viii

Figure 2.15 Calculated temperature from 2-D model for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan off ................................................25 Figure 2.16 Cooling flux produced by thermoelectric cooler during cooling from 94 to 54 °C, 13.8 V Input, cooling fan on ...............................26 Figure 2.17 2-D calculated temperature for top of the TEC superimposed on graph of actual temperature for top of TEC for 13.8 V input and cooling fan on ...............................................................................................27 Figure 2.18 Cooling flux produced by TEC during cooling from 94 to 54 °C, 13.8 V input, cooling fan off ....................................................28 Figure 2.19 Techne thermocycler analyzed in this section ..............................................31 Figure 2.20 Aluminum heating block mounted on top of TECs in Techne thermocycler ......................................................................32 Figure 2.21 Simulated PCR cycle using commercial Peltier thermocycler .....................33 Figure 2.22 PCR reaction results for extension time test, 40 s elongation time ..............36 Figure 2.23 PCR reaction results for extension time test, 30 s elongation time ..............37 Figure 2.24 PCR reaction results for extension time test, 25 s elongation time ..............38 Figure 2.25 PCR reaction results for extension time test, 30 s elongation time with 15 s at denaturation and annealing steps.....................39 Figure 2.26 Glass capillaries run with a 1/4000 volumetric concentration of ethidium bromide......................................................................................40

Figure 3.1 Convective thermocycler setup #1................................................................42

Figure 3.2 Heater mounted within vent pipe..................................................................43 Figure 3.3 Double fan convective thermocycler setup...................................................43 Figure 3.4 Heater element within double fan convective thermocycle..........................44 Figure 3.5 Simulated PCR cycle using the single fan convective thermocycler............46 Figure 3.6 Simulated PCR cycle using the double fan convective thermocycler ..........46

ix

Figure 3.7 Cross section of single fan convective thermocycler showing mixing walls located within vent pipe............................................52 Figure 3.8 Radial temperature measurements for single fan convective thermocycler with heater and fan at duty cyle = 1........................................48 Figure 3.9 Schematic of control volume used for air velocity determination................50 Figure 3.10 200 µL PCR tube with 10 µL of water used in convective analysis.............55 Figure 3.11 2-D model used for convective analysis of 200 µL polypropylene tube filled with 10 µL of water .....................................................................58 Figure 3.12 Experimental results for single fan convective thermocycler with 200 µL tube filled with 10 µL of water placed inside vent pipe ..........59 Figure 3.13 2-D model correlation with experimental results for single fan convective thermocycler with 200 µL PCR tube containing 10 µL of water during heating from 52 – 94 °C ...........................................60 Figure 3.14 2-D model correlation with experimental results for single fan convective thermocycler with 200 µL PCR tube during cooling from 94 – 52 °C ...............................................................................61 Figure 3.15 Experimental results for double fan convective thermocycler with 200 µL polypropylene tube filled with 10 µL of water placed inside vent pipe..................................................................................62 Figure 3.16 Lumped capacitance model with experimental results for double fan convective thermocycler with 200 µL PCR tube during heating from 54–94 °C ..............................................................63 Figure 3.17 Lumped capacitance model with experimental results for double fan convective thermocycler with 200 µL PCR tube during cooling from 94–54 °C .................................................................................65 Figure 3.18 Plot of time constant versus air velocity for cylindrical configuration with constant diameter, length, wall thickness, surface temperature, and air temperature......................................................67 Figure 3.19 Plot of time constant versus diameter for cylindrical polypropylene configuration ........................................................................68 Figure 3.20 Lumped capacitance model of reaction volume temperature for glass capillary of 890 µm outside diameter with 5 µL of water inside ...................................................................................................70

x

Figure 3.21 Plot of time constant versus diameter for spherical configuration with constant air velocity, and wall thickness ..............................................72 Figure 3.22 Biot number versus wall thickness for a poly propylene capillary of outside diameter 890 µm ...........................................................75 Figure 3.23 Biot number versus wall thickness for a glass capillary of outside diameter

890 µm ..........................................................................................................76 Figure 4.1 Heater with integrated well as designed by El-Ali et al................................78 Figure 4.2 PCR reaction vessel designed by Zhao et al. ................................................80 Figure 4.3 Microheaters with integrated reaction vessels as designed by Zou et al. .....81 Figure 4.4 Flip chip design process used by Zou et al. for fabrication of PCR

microheaters..................................................................................................82 Figure 4.5 Schematic of 2-D model used in analysis of heating power.........................86 Figure 4.6 Response of 3.2 mm diameter polycarbonate chamber of thickness 2 mm

with 10 µL of water in it and a heat flux of 20 W/cm2 ................................86 Figure 4.7 Response of 5 mm diameter polycarbonate chamber of depth 2 mm with a 10 µL sample of water in it and a heat flux of 20 W/cm2..................88 Figure 4.8 Schematic of 5 mm heater.............................................................................91 Figure 4.9 Exploded view of 5 mm heater with some elements deleted for clarity.......92 Figure 4.10 Schematic of 2.5 mm heater..........................................................................93 Figure 4.11 Exploded view of 2.5 mm heater ..................................................................94 Figure 4.12 Gold lead mask for 5 mm heater...................................................................95 Figure 4.13 Gold lead mask for 2.5 mm heater................................................................95 Figure 4.14 Circuit schematic for microheater control ....................................................96 Figure 4.15 Schematic of circuit used to measure temperature of platinum temperature sensor ........................................................................................98 Figure 4.16 Simplified fabrication sequence for microheaters ......................................101 Figure 4.17 Effect of chlorobenzene on photoresist development.................................102

xi

Figure 4.18 Delamination of heater elements from soda lime glass substrate ...............104 Figure 4.19 Heater elements from a heater with no discontinuities...............................105 Figure 4.20 Comparison of 30 µm heater elements with 50 µm temperature sensor ....105 Figure 4.21 Points of resistance measurement for 5 mm microheater ...........................107 Figure 4.22 Temperature versus resistance with linear fit for microheater....................108 Figure 4.23 Temperature versus resistance graph with linear fit for microheater

temperature sensor ......................................................................................109 Figure 4.24 Comparison of initial calibration of temperature sensor and after 1 week of

use ...............................................................................................................110 Figure 4.25 Microheater test fixture...............................................................................111 Figure 4.26 Microheater test fixture with microheater mounted on heat sink ...............112 Figure 4.27 Heating and cooling characteristics of microheater with cooling fan, no heat sink, 12 V input ..............................................................................113 Figure 4.28 Heating and cooling characteristics of microheater with cooling fan, heat sink, 12 V input ...................................................................................114 Figure 5.1 Silicon reaction chamber fabricated by Daniel et al. ..................................116 Figure 5.2 Simplified fabrication sequence used by Daniel et al. ................................117 Figure 5.3 Polycarbonate chamber fabricated by Yang et al. ......................................120 Figure 5.4 2-D heat transfer model for polycarbonate chamber ..................................124 Figure 5.5 2-D conduction model of 5 mm square polycarbonate chamber on 5 mm

square microheater ......................................................................................125 Figure 5.6 2-D conduction model of 3 mm square polycarbonate chamber on 5 mm

square microheater ......................................................................................126 Figure 5.7 2-D conduction model of 3 mm square silicon chamber on 5 mm square

microheater .................................................................................................127 Figure 5.8 Schematic of 800 µm capillary on microheater ..........................................129

xii

Figure 5.9 Polycarbonate reaction vessel thermally bonded from three 250 µm thick sheets...........................................................................................................130

Figure 5.10 Predicted pressure rise inside closed reaction vessel with water and 33% of

the volume filled with air............................................................................132 Figure 5.11 Predicted bubble volume inside closed reaction vessel with water and 33%

of the volume filled with air........................................................................133 Figure 5.12 Polycarbonate chamber fabricated from three 250 µm thick sections........134 Figure 5.13 Countersunk polycarbonate chambers created with router bit....................135 Figure 5.14 Bubble formation in untreated polycarbonate chambers covered with

mineral oil ...................................................................................................137 Figure 5.15 2 mm thick polycarbontate after O2 plasma etch at A)100 W, B) 200 W, C)

250 W..........................................................................................................138 Figure 5.16 PCR cycle showing 5 mm square microfabricated heater and temperature

inside a 400 µm inside diameter glass capillary filled with ethylene glycol ............................................................................................140 Figure 5.17 PCR cycle showing 5 mm square microfabricated heater and temperature

inside a 5 mm square polycarbonate reaction volume filled with ethylene glycol....................................................................................141 Figure 5.18 Simulated PCR cycle showing 5 mm square microfabricated heater and

temperature inside a 3 mm square polycarbonate reaction volume filled with ethylene glycol ............................................................................................142

Figure 5.19 Simulated PCR cycle showing 5 mm square microfabricated heater and

temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick top and bottom section filled with ethylene glycol .....................144


temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick bottom section and a 250 µm thick top section filled with ethylene glycol...........................................................................................................145

xiii

SYMBOLS

A area, m2 Bi Biot number

c specific heat, J/kg K cp specific heat

D diameter, m at constant pressure, J/kg K

f friction factor; fraction G area/length of

h convection coefficient, W/m2 K; thermoelectric device (cm)

enthalpy, J/kg

I electrical current, A k thermal conductivity, W/m K

L length, m m mass, kg

N number Nu Nusselt number

P power, W; Pressure, N/m2 Pr Prandtl number

q heat transfer rate, W "q heat flux, W/m2

R electrical resistance, Ω; universal gas Re Reynolds number

constant divided by the molecular

weight, kJ/kg K

T temperature,°C, K t time, s; thickness, m

V voltage, V; volume, m3; Velocity, m/s W percent on time; work, J

w width, m x displacement, m; quality

Greek letters

α Seebeck coefficient, V/K ∆ change in

υ specific volume, m3/kg μ viscosity, kg/s m

ρ mass density, kg/m3; electrical τ time constant, s

resistivity, Ω-m

xiv

Subscripts

air refers to air ave average

B beginning c during cooling; cold

D diameter side; characteristic

E end h during heating, hot side

i initial condition; inlet condition l liquid

max maximum min minimum

n point n o outlet condition

P parallel p polypropylene

RT thermometer resistance S series

s surface; solid t thermal; thermometer

tot total v vapor

w water wl liquid water

wv water vapor ∞ free stream conditions

Overmarks

. per unit time, /s surface average conditions,

time mean

1

1. INTRODUCTION TO PCR AND THERMOCYCLERS

PCR is a process by which a small amount of DNA is amplified many times to

yield an easily detectable amount. This process is widely used for the detection of

bacterial pathogens for biodefense (Draghici, Khatri et al. 2005; Hindson, Makarewicz et

al. 2005). PCR is also utilized in other applications such as basic research, criminal

identification, and disease detection in humans (Aryee, Bailey et al. 2005; Gill 2005).

DNA, primers that dictate what part of the DNA is to be copied, nucleotides that form the

building blocks for copying, buffer, polymerase that is the machine for copying, and

magnesium ions are needed for the reaction (Cadoret, Rousseau et al. 2005). Three

temperature levels are required in a typical PCR reaction. The first step involves heating

to around 94 °C. The double stranded DNA is denatured, or split into two single strands

during this step. The second step involves cooling to approximately 54°C to allow the

primers to attach to the single-stranded DNA. The primers are what determine which

section of DNA will be copied. They are designed such that they straddle a

predetermined specific site on the DNA based on the sequence of the DNA. The third

step is performed at around 72 °C. The actual copying of the DNA is performed by the

DNA polymerase during this step. The DNA polymerase attaches to the primer site and

copies DNA using the nucleotides that are in solution as building blocks. This cycle is

repeated approximately 30 times (Nam, Srinivasan et al. 2005). The steps in the PCR

cycle are illustrated graphically in Figure 1.1. The temperature profile of a PCR cycle is

shown in Figure 1.2.

2

Figure 1.1 One cycle of the PCR process (Vierstraete 2004)

45

50

55

60

65

70

75

80

85

90

95

100

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Time (seconds)

Tem

pera

ture

(°C

)

Figure 1.2 A PCR cycle showing plateaus at 54, 72, and 94 °C

3

A typical PCR reaction in a commercial thermocycler takes between 30 minutes

(LightCycler) and 2 hours (typical Peltier systems). This time duration is acceptable for

some applications, however, many other applications exist where a rapid reaction is

desirable such as clinical applications where treatment depends on the result. The

primary objective of this thesis was to investigate system alternatives for high speed

PCR. A secondary objective was to minimize power consumption to allow eventual

portable operation with battery power. A series of alternative designs were tested in this

effort and each was found to have a set of unique and useful characteristics.

The cycle time can vary based on the situation. The cycle time is strongly

dependent upon the characteristics of the thermocycler. If the thermocycler takes a long

time to reach the setpoint, then this must be accommodated. If heat conduction or

convection is slow through the reaction chamber in spite of a fast heater, then the cycle

time would also be longer. PCR reactions run in this lab in commercially available

Peltier type thermocyclers are usually run for two hours. This type of thermocycler has

an aluminum heating block that weighs 41 g. This block takes time to heat up and slows

the heating of the actual reaction volume. The commercially available convective

LightCycler has been reported to produce PCR results in as little as 20 minutes (Nitsche,

Steur et al. 1999). One reason PCR occurs so rapidly in this device is that the reaction

volume heats up very quickly. Another factor that influences the speed of a PCR reaction

is the length of DNA strand that is to be copied. Shorter strands take less time to copy

then longer strands. The insertion rate for nucleotides during primer extension (72 °C) is

50 base pairs /s (Scheinert, Behrens et al. 2005). A base pair (bp) is simply one link

between the two single strands of DNA. If the DNA fragment to be copied is only 100

4

bp then it would require only two seconds at 72 °C to be copied. If the DNA fragment

was 400 bp then it would require 8 seconds to be copied. The type of DNA polymerase

that is used can significantly alter the reaction time. Hot Start TAQ DNA polymerase

requires 15 minutes of heating at 95 °C to become effective. This type of polymerase is

commonly used in PCR reactions because it reduces the occurrence of nonspecific

amplicons (Kermekchiev, Tzekov et al. 2003). The term amplicon refers to the DNA

segments produced from the PCR reaction. Other types of DNA polymerase may not be

as specific, but they do not require this 15 minute heating time. In conclusion, the cycle

time is influenced by the thermocycler, the reaction volume, the length of the DNA

segment to be copied, and the type of DNA polymerase used.

The amplified DNA may be detected in a number of ways. One way is by taking

the sample and moving it through a gel with the application of an electric field. Smaller

DNA segments migrate through the gel more quickly while larger DNA fragments move

more slowly. The DNA fragment is detected by determining how far it migrates through

the gel. The distance migrated through the gel is revealed by illuminating the DNA under

a UV light source. The UV light causes fluorescent dye that is added to the DNA to

fluoresce, thus enabling one to see the location of the DNA fragment. Another method of

detection involves the use of probes that have the complementary sequence to the single

strand of DNA that is being copied. If probe locations are occupied with double stranded

DNA, then it is known that PCR occurred on that particular piece of DNA (Lane,

Everman et al. 2004).

PCR requires that the reaction volume undergo changes in temperature, therefore

a thermal cycler is needed to perform this operation. Thermal cyclers come in a variety

5

of shapes and sizes, but generally may be broken down into two branches: convective

thermocyclers and conductive thermal cyclers with other types being reported such as

microwave assisted (Fermer, Nilsson et al. 2003) and infrared (Lee, Tsai et al. 2004).

Convective thermocyclers operate on the principal of flowing air past a reaction vessel in

order to increase or decrease the temperature of the reaction vessel. These units consist

of a reaction chamber, housing, heater, fan, and control system (Wittwer, Fillmore et al.

1989). The other type of thermocycler relies on conduction from a heated plate to drive

the reaction. The heated plate could take on a variety of shapes and sizes varying from a

thermoelectric cooler to a microfabricated heater. The purpose of this thesis is to

investigate the properties of these different types of thermocyclers.

1.2 PCR reaction and detection reagents

PCR was run many times to determine the efficacy of the thermocyclers tested.

Several reagents are necessary to run the reaction. The DNA was genomic DNA from

non-infectious B. anthracis (Sterne strain) which is the strain used to make vaccine. Hot

start TAQ DNA polymerase, 10 X buffer solution, and 25 mM MgCl2 solution supplied

by Qiagen were used in all PCR runs. The dNTPs, which are the building blocks the

DNA polymerase uses to copy the target DNA, were obtained as a 10 mM solution from

Fisher Scientific. Primers that determine which section of DNA is to be copied were

designed and ordered from Operon. The gene amplified was the piplc gene of B.

anthacis. Table 1.1 shows the exact sequence of the primers used and the amplicon

length that was expected based on the designated primers.

6

Table 1.1 Primer sequences used in amplification of B. anthracis

Primer Sequence Amplicon length piplc-R ATA TTT ATC ACA TTT GTA TTT GCT TTA CAT GA 879 piplc-F CTT CTT GAT ACA ATA ATG GTG ACC ACT piplc-R2 ATG AAA CGA TGA ACA ATA GTG AGG A 216

The R and F suffixes at the end of the primer name in the first column of Table 1.1

designate whether the primer was a forward primer or a reverse primer. Forward primers

attach to one end of the single stranded DNA segment to be copied while reverse primers

attach to the other end of the complementary strand. Copying proceeds as shown in

Figure 1.1. The first two primers listed in Table 1.1 were used to amplify a segment of

DNA 879 bp long. The third row of Table 1.1 is the reverse primer that was used with

the forward primer to produce an amplicon of length 216 bp. Distilled water was used in

all reactions. Bovine serum albumin was also added to the reactions. Bovine serum

albumin helps to coat the reaction vessel and stop protein adsorption to the surface (Liu,

Rauch et al. 2002).

One method used to detect the PCR products was gel electrophoresis. A QS710

horizontal gel electrophoresis unit was obtained from Fisher Scientific. The power

supply used with the gel electrophoresis unit was a Buchler DC power supply, model

31153. A Kodak EDA 290 camera stand was mounted on top of a Foto Prep I by

Fotodyne UV light source to take pictures of the gel. Kodak 1D version 3.6 was the

software used to take the pictures. Omni Pur PCR plus agarose was used to separate the

PCR products. The buffer solution used in the electrophoresis unit was TBE Omni Pur

10 X powder. Ethidium bromide was used to attach to the double stranded DNA during

gel electrophoresis. Ethidium bromide fluoresces under UV light when attached to

7

double stranded DNA. Exact gene 100 bp PCR DNA ladder was used to determine the

length of PCR amplicons.

Another method used to detect the occurrence of PCR was by adding ethidium

bromide to the PCR solution at a concentration of 1:4000 by volume. This method

worked well for illuminating PCR product in glass capillaries and sped up the PCR

process by skipping the time consuming gel electrophoresis step.

1.3 Thermocycler metrics

Metrics that will be used to discuss the performance of the thermocyclers are

heating rate, cooling rate, average heating rate, average cooling rate, and power

consumption. The maximum heating rate is defined as the maximum rate of change of

temperature when the heater is active during heating within the temperature range of a

typical PCR reaction (54 to 94 °C). The heating rate is based on the surface temperature

of the heater, or in the case of the convective thermocycler, the air temperature.

max

.

dtdTT h = (1.1)

where T is the temperature and t represents time. The cooling rate is defined as the

maximum rate of decrease in temperature during cooling from 94 to 54 °C.

min

.

dtdTT c = (1.2)

The average cooling and heating rate are defined as follows

12

12_

ttTTTh −

−= ,

12

12_

ttTTTc −

−= (1.3)

8

Figure 1.3 Schematic of PCR cycle with dashed

areas representing times included in the average

power over cycle calculation

The average rates are influenced by the control system implemented to control the

temperature of the heater, unlike the maximum heating and cooling rates, which are

maximum values calculated when the heater was less inhibited by the control system.

The average heating rate would be much faster if the heater was on at full power during

the entire interval between T2 and T1. This is not the case, however, because the control

system intervenes before the heater temperature reaches the set point in order to avoid

overshoot.

The energy consumption can be reported in several ways. The maximum energy

consumption per unit time during heating and cooling is one of those ways. During the

heating stages where the PCR reaction increases in temperature, the heater is on at full

power. This means that the power required is given by

VIP = (1.4)

where P is the power, V is the voltage, and I is current draw of the device.

The power requirement

varies throughout the cycle, so the

average power over a cycle can be

used to characterize the system. The

average power over a cycle includes

the time required for heating and

cooling as well as five seconds of

each plateau period as shown in

Figure 1.3.

9

The average power is defined in this way in order to compare different heaters.

All of the heaters tested had at least a five second plateau period, so this method would

compare them with the same standard. The average power over a cycle could be

increased or decreased if the energy of a very long plateau was included in the

calculation. The average power over a cycle is given by the following equation

tPN

P n

N

cycleave ∑=0

/1 (1.5)

where Pave/cycle is the average power over a cycle, N is the number of seconds in the cycle,

t is the time of each period, and Pn is the power used by the system during period n.

Some of the systems described in this thesis were controlled using a pulse width

modulation scheme where the power was applied during some fraction of each control

period (period used here was 1 sec). This control scheme is referred to as pulse width

modulation (PWM). In this case, Pn is given by

VIWPn = (1.6)

where W is the time per period that the heater was on.

10

2. TESTING OF PELTIER THERMOCYCLER FOR PCR

A thermoelectric cooler (TEC) was tested to determine its efficacy for performing

PCR. The thermoelectric cooler was mounted on an aluminum heat sink with an attached

cooling fan. The TEC was tested for power consumption, heating and cooling rate, and

heating and cooling flux. A commercially available thermocycler using four

thermoelectric coolers with an aluminum block to hold reaction vessels was also tested to

determine the same properties.

2.1 Testing of an independent TEC, apparatus and procedures

A single TEC was tested using the setup shown in Figure 2.1. A Mastech power

supply was used as the voltage source. A Type T thermocouple was used to monitor the

temperature of the surface of the heater. The InstruNet was used as the data acquisition

and controller for the setup.

Figure 2.1 Thermoelectric thermocycler setup

InstruNet model 100 analog/digital input/output system

Omega solid state relay part no. SSRDC100VDC12

Melcor thermoelectric cooler, model HT2-12-30-T2

(50 x 50 x 7) mm finned heat sink with 12 VDC fan

Copper-constantan (type T) thermocouple

Mastech DC power supply part no. HY1803D

11

The thermoelectric cooler (TEC) was controlled with a Visual Basic program

interfaced with the InstruNet model 100 analog/digital input/output system. The TEC

was controlled with a pulse width modulation scheme with a pulse frequency of 1 Hz.

The program calculated the appropriate pulse width and controlled the relays accordingly

via the InstruNet interface. Four solid state relays were utilized in an H-bridge

configuration allowing the direction of the current to be switched based on whether the

TEC was in heating or cooling mode as shown in Figure 2.2.

Figure 2.2 H-bridge configuration used in the control of the TEC. The components

labeled SSRi are solid state relays. Arrow on figure shows current flow in heating mode.

A forward current flows through the TEC when SSR1 and SSR3 are opened via a signal

from InstruNet (SSR2 and SSR4 remain closed). This is the heating mode of operation.

The TEC was put in cooling mode by closing SSR1 and SSR3 and opening SSR2 and

SSR4. This configuration allowed current to flow through the TEC in the opposite

direction.

12

The thermoelectric cooler was placed on the heat sink as shown in Figure 2.3.

The bottom of the TEC was coated with synthetic grease to aid in heat transfer between

the heat sink and the TEC. A glass slide, representing an in situ PCR configuration, was

mounted on top of the TEC and held in place with two metal strips. The thermocouple

was attached to the slide and held in place with a piece of tape. The purpose of this

thermocouple was to control the temperature of the slide. Another thermocouple was

taped to the heat sink. In some tests, thermocouples were attached to the top and bottom

surfaces of the TEC with glue in order to monitor the temperature of those two surfaces.

Voltage and current readings were obtained from the digital display on the power supply

(Mastech HY1803D). The TEC was tested over the range 9 – 13.8 VDC to assess the

effect on power consumption.

Figure 2.3 Side view of TEC setup

The data acquisition program recorded data once per second. Data recorded

included the temperature of the slide and top of TEC, the pulse width, time, and set point

TEC

Cooling fan

Heat sink

Glass slide

Metal compression strip

Compression screw Grease between slide and TEC and between TEC and heat sink

13

temperature (set point temperature varied during the run according to a programmed

input function). Using the recorded data, heating rate, cooling rate, and power

consumption were calculated. Only the power used by the heater was included in the

calculations for power consumption. Power consumed by the heat sink cooling fan, the

computer, and power supply were not included. The heating and cooling rates were

calculated as described in the thermocycler metrics section.

2.2 Independent TEC testing results

The TEC was tested with two different configurations to determine its behavior

under different conditions. The first condition was with the cooling fan on the heat sink

on. The fan was run continuously and was never turned off. The second condition was

with the fan on the heat sink off at all times.

The maximum heating rate for the TEC was 5.78 K/s with an input voltage of

13.8 V without the heat sink fan turned on as shown in Figure 2.4. The top of TEC curve

in Figure 2.4 also shows that the average heating rate for heating between 72 and 94 °C

was 2.0 K/s while the average heating rate between 54 and 72 °C was 1.4 K/s. The

maximum cooling rate for this setup was 5.2 K/s. The average cooling rate between 94

and 54 °C was 4.3 K/s.

14

253035404550556065707580859095

100

0 15 30 45 60 75 90 105 120 135 150 165 180

Time (seconds)

Tem

pera

ture

(°C

)

SetpointTop of TECBottom of TEC

Figure 2.4 Simulated PCR cycle for TEC without cooling fan , 13.8 V input

The TEC was next tested with the cooling fan on the heat sink on. The top of

TEC trace in Figure 2.5 was used to calculate the maximum heating and cooling rates as

well as the average rates. The maximum heating rate was 5.2 K/s. The average heating

rate between 54 and 72 °C was 1.8 K/s while the average heating rate between 72 and

94 °C was 1.3 K/s. The maximum cooling rate of the TEC with cooling fan was 6.6 K/s.

The average cooling rate was 5.8 K/s.

253035404550556065707580859095

100

0 15 30 45 60 75 90 105 120 135 150 165 180Time (seconds)

Tem

pera

ture

(°C

)

SetpointTop of TECBottom of TEC

Figure 2.5 Simulated PCR cycle for TEC with cooling fan on, 13.8V input

15

The TEC was next tested to determine its behavior at steady state. The TEC was

allowed to reach steady state while receiving full power (duty cycle = 1) and the results

were recorded. The TEC was powered with 10 V during this testing. The results for the

case where the TEC was allowed to reach steady state in heating mode with no cooling

fan are shown in Figure 2.6. Heating mode means that the top of the TEC was hot while

the side against the heat sink was cold.

Figure 2.6 Steady state heating mode of the TEC without cooling fan and 10.0 V input

The final temperature differential between hot and cold side was 55 K for the case

where the TEC was run in heating mode at steady state. The cold side temperature grew

to approximately 70 °C by the end of the test.

The TEC was run in heating mode at steady state with the cooling fan on. The

results of the test with a 10.0 V input are shown in Figure 2.7. The temperature of the

16

cool side was limited to 40 °C. The final temperature differential between hot and cold

side was 63 K as shown in Figure 2.7.

Figure 2.7 Steady state heating mode of the TEC with cooling fan and 10.0 V input

The TEC was next operated at steady state conditions in cooling mode with a 13.8

V input. The top of the TEC was the cold side during cooling while the hot side was

against the heat sink. The TEC was supplied with full power during this test and no

control system was used. The results were recorded until the system reached steady state.

Figure 2.8 Steady state cooling mode of the TEC without cooling fan and 13.8 V input

17

The steady state cooling mode of the TEC is shown in Figure 2.8. The final

temperature differential between hot and cold side was 42 K. It is noted that the cold side

trace in Figure 2.8 does not appear to be at steady state because this test was terminated

early. However, with the exception of an accurate measurement of the steady state

temperature difference, this test demonstrated the key aspects of interest.

A similar test with the cooling fan on is shown in Figure 2.9. The test shows the

cooling characteristics of the TEC from 94 °C. The test was run until steady state was

reached, as can be seen from the graph. The cold side of the TEC decreased to almost 0

°C. A temperature differential between top and bottom of approximately 50 K was

established.

Figure 2.9 Steady state cooling mode of the TEC with cooling fan and 13.8 V input

18

2.2.1 Heating flux characteristics of TEC

Modeling was done to better understand the heat flux characteristics of the TEC.

The data from the thermocouple placed on top of the TEC was used to calculate the heat

flux as a function of time during the heating and cooling modes of the TEC. The heating

of the top surface of the TEC which is made of aluminum oxide, also known as alumina,

is a transient problem, therefore transient methods must be used. The lumped

capacitance model is useful because it does not consider heat conduction within the solid.

The lumped capacitance method utilizes an energy balance around the solid. This

approach allows one to solve the transient problem easily. The lumped capacitance is

most accurate in situations where the Biot number is much less than one. If the Biot

number is much less than one, then convection from the surface occurs much more

readily than conduction within the solid, therefore conduction within the solid may be

ignored (Incropera and DeWitt 1996). Radiation effects were ignored in this case and the

convection coefficient was assumed to be constant with time. The lumped capacitance

method was selected to model this transient problem because the Biot number in this case

is 1.26 x 10-4, assuming that the convection coefficient is 5 W/m2K, the characteristic

length is 1 mm, and aluminum oxide has the properties shown in Table 2.1.

Table 2.1 Properties of aluminum oxide used in the analysis (Accuratus 2005)

Property Aluminum OxideThermal Conductivity (W/m K) 25Density (kg/m^3) 3720Specific Heat (J/kg K) 880

The use of the lumped capacitance method greatly simplifies the problem. For the

conditions of this problem, the lumped capacitance model has an analytical solution as

(Incropera and DeWitt 1996)

19

))exp(1(/)exp( atTTabat

TTTT

ii

−−−

+−=−−

∞∞

∞ (2.1)

where a and b are given by

VchA

a S

ρ= and

VcAq

b ss

ρ

"

= (2.2)

Equations 2.1 and 2.2 were solved implicitly for the heat flux using temperature

data from the test. This was possible because the temperature at the beginning and end of

each second was known and could be inserted into the equation for T and Ti. All of the

other variables are known, except for qs”. The properties used in the equation are the

alumina properties listed in Table 2.1. The other variables were

As=0.0009 m2 (0.030 x 0.030) m2, surface dimensions of alumina

V=9 x 10-7 m3 (0.030 x 0.030 x 0.001) m3, volume of alumina

The heat flux produced by the TEC during heating from 72 to 94 °C with cooling

fan as calculated with the lumped capacitance method is graphed in Figure 2.10 along

with the corresponding temperature of the top and bottom of the TEC. The TEC was on

at full power for the first five seconds of the cycle meaning that the duty cycle was 100%,

after that the duty cycle was controlled with the control system. The maximum heat flux

within the 72 to 95 °C range with a 13.8 V input and cooling fan on was 1.23 W/cm2.

The maximum heat flux occurred at t = 3 seconds. The heat flux decreased for the next

two seconds despite the fact that the TEC was on at full power.

20

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18Time (seconds)

q" (W

/cm

^2)

253035404550556065707580859095100

Tem

pera

ture

(°C

)

q"Top of TEC temperatureBottom of TEC temperature

Figure 2.10 Heat flux produced by thermoelectric cooler during heating from 72 to

94 °C and corresponding top and bottom of TEC temperature with 13.8 V input, cooling

fan.

The reason the heat flux increased to a maximum and then began to decrease,

despite the fact that the TEC was on at full power can be attributed to the characteristics

of the TEC. The relevant governing equations of a TEC are given by (Melcor 2005)

)2

(22

TGkG

IITNq Cc Δ−−=ρα (2.3)

)(2 TGINV Δ+= αρ (2.4)

ch qWq += & (2.5)

Where cq is the heat pumped at the cold surface of the TEC, hq is the heat pumped at the

hot surface of the TEC, N is the number of thermocouples in the TEC, α is the Seebeck

21

coefficient, I is the electrical current consumed by the TEC, Tc is the cold side

temperature, ρ is the resistivity of the bismuth telluride which makes up the TEC

junctions, G is the area divided by the length of a thermoelectric element, k is the thermal

conductivity of the bismuth telluride, ΔT is the difference in temperature between the hot

side and the cold side of the TEC, V is the voltage supplied to the TEC, and W& is the

power supplied to the TEC via the power supply.

These equations may be used to estimate the performance of the TEC and

determine trends in its performance. The calculated heating and cooling fluxes as well as

the current for the TEC were determined by inserting the recorded values for the hot and

cold side temperatures into equations 2.3-2.5 with the supply voltage value of 13.8 V.

The material properties of the bismuth telluride are variable with temperature and were

obtained from the manufacturer (Melcor 2005). The results of the analysis are shown in

Figure 2.11.

25

30

35

40

45

50

55

60

65

70

75

80

85

90

95

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Time (seconds)

Tem

pera

ture

(°C

)

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

q" (W

/cm

^2)

ThTcInput powerCooling fluxHeating flux

Figure 2.11 Calculated Performance of TEC using hot and cold side temperatures from

experiment

22

The hot and cold side temperatures are plotted on the left axis while the heating flux,

cooling flux, and input power per unit area are plotted against the right axis. The results

indicate that as the difference in temperature between the hot and cold side of the TEC

increases, the heat flux, input power, and cooling flux all decrease. The reason for this is

clear looking at Equations 2.3-2.5. The value of qc decreases as the temperature

difference increases because Tc decreases and ΔT increases. The power decreases also

because the current decreases as the difference in temperature between the hot and cold

side increases.

The reason that the heat flux increases and then decreases, even though the TEC

was operating at full power is because as the temperature difference between top and

bottom increases, the heating flux decreases as the above analysis shows. The initial

ramp of the heating flux may be attributed to transient operation of the device while

decreases of the heat flux after the first five seconds may be attributed to the fact that the

input power to the device was reduced by the control system.

The maximum heat flux calculated by the Melcor model was 3 times greater than

the actual heat flux. Melcor recommends that an insulating boundary be placed around

the TEC. The TEC was not insulated in this case leading to losses to convection. The

error could also be caused by the model not relating to transient conditions well. The

model was useful for looking at trends in the heating and cooling flux, however.

The heat flux information obtained from the above analysis was then applied to

the bottom surface of a system model using a two dimensional finite element heat transfer

program called Finite Element Heat Transfer (FEHT). The nodal temperatures on the top

surface of the model were then compared with the experimental data in order to verify

23

that the heat flux information was correct. The 2-D model geometry is shown in Figure

2.12. The alumina plate is 30 x 30 x 1 mm.

Figure 2.12 2-D geometry used for cooling and heating model. All dimensions in mm.

The results, superimposed on the graph for the experimental results, are shown in Figure

2.13. The calculated surface temperatures differ by at most 1.7 K from the measured

values. This result indicates that the lumped capacitance method is a valid method for

calculating the heat flux for transient heating.

253035404550556065707580859095

100

0 2 4 6 8 10 12 14 16 18

Time (seconds)

Tem

pera

ture

(°C

)

2-D modelTop of TEC

Figure 2.13 Calculated temperature for top of the TEC superimposed on graph of actual

temperature for top of TEC for 13.8 V input and cooling fan on.

Time dependent heat flux BC

Aluminum oxide

Three other outer surfaces convective heat transfer coefficient of 5 W/m2 K

24

A similar calculation was performed to analyze the heating characteristics of the

TEC with the heat sink cooling fan off and a 13.8 V input. The calculated heat flux is

shown in Figure 2.14. The maximum heat flux for the TEC during heating from 72 to

94 °C was 1.92 W/cm2 with the heat sink fan off and a 13.8 V input.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18Time (seconds)

q" (W

/cm

^2)

253035404550556065707580859095100

Tem

pera

ture

(°C

)

q"Top of TEC temperatureBottom of TEC temperature

Figure 2.14 Heat flux produced by thermoelectric cooler during heating from 72 to

94 °C with 13.8 V input, cooling fan off

The heat flux for the case without the cooling fan on and a 13.8 V input reacts

differently then the case where the cooling fan was on. The heat flux for the case without

cooling fan increases for the first three seconds. The TEC was being operated with a

100 % duty cycle during this time. The TEC was pulsed after the three second point.

This differs from the case with the cooling fan on, where the heat flux increased and then

decreased, despite the fact that the TEC was on at full power. The temperature of the

25

heat sink in the case with no cooling fan was 30 K warmer than the case with heat sink

fan on. The fact that the heat sink was warmer and the temperature difference between

top and bottom of TEC was less, created a higher heat flux and made it possible for the

heat flux to not decrease when on at full power as explained in the discussion of Figure

2.10.

The heat flux calculated in Figure 2.14 was input into the 2-D model and the

simulation was run to predict temperature. The calculated values are superimposed on

the temperature data in Figure 2.15. The maximum error between the model and actual

temperatures was 1.8 K.

253035404550556065707580859095

100

0 2 4 6 8 10 12 14 16 18Time (seconds)

Tem

pera

ture

(°C

)

2-D model

Top of TEC

Figure 2.15 Calculated temperature from 2-D model for top of the TEC superimposed

on graph of actual temperature for top of TEC for 13.8 V input and cooling fan off.

26

2.2.2 Cooling flux characteristics of TEC

The cooling flux of the TEC was analyzed with the same method as used to

determine the heating flux of the TEC. The 2-D model and the properties for the material

were the same. The first case analyzed was the case with the cooling fan. The results

from the test are shown in Figure 2.16.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0 1 2 3 4 5 6 7 8 9 10

Time (seconds)

qc"

(W/c

m^2

)

253035404550556065707580859095100

Tem

pera

ture

(°C

)

qc"Top of TEC temperatureBottom of TEC temperature

Figure 2.16 Cooling flux produced by thermoelectric cooler during cooling from 94 to

54 °C, 13.8 V Input, cooling fan on

Figure 2.16 shows the cooling flux characteristics of the TEC as calculated with

the lumped capacitance method. The maximum cooling flux produced by the TEC was

2.1 W/cm2 for the case with cooling fan on and a 13.8 V input. The TEC was on at full

power through t= 6 s. The cooling flux increases, reaches a peak, and then decreases as

the temperature difference between the top and bottom of the TEC decreases. This is the

27

exact opposite behavior observed in the case where the TEC was in heating mode. The

cooling flux increased as the difference between the top and bottom of the TEC decreased

in the heating mode case. The equations governing the behavior of the TEC may once

again be used to understand the operation of the TEC. In the heating mode case, the

value of ΔT increases because ΔT = Th-Tc. The value of ΔT is increasing in the cooling

mode case also because the difference in temperatures at t = 0 seconds is a negative

number. As time proceeds, the value of ΔT becomes less negative, so it is indeed

increasing. Thus, what appears to be a contradiction between the heating mode and the

cooling mode case is actually the same case, so it makes sense that the cooling flux

decreases as the hot and cold side temperatures get close to one another.

The tabulated cooling flux results were next put into the 2-D model as a time

dependent heat flux on the bottom surface of the alumina. The results of that analysis are


253035404550556065707580859095

100

0 1 2 3 4 5 6 7 8 9 10Time (seconds)

Tem

pera

ture

(°C

)

Top of TEC 2-D model

Figure 2.17 2-D calculated temperature for top of the TEC superimposed on graph of

actual temperature for top of TEC for 13.8 V input and cooling fan on

28

The maximum error between the measured values of temperature and the

calculated temperature was 2.2 K, indicating that the lumped capacitance model is an

acceptable model for this system.

A similar procedure was used to calculate the cooling flux produced by the TEC

during cooling with the cooling fan off. The results are shown in Figure 2.18.

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0 1 2 3 4 5 6 7 8 9 10

Time (seconds)

qc"

(W/c

m^2

)

253035404550556065707580859095100

Tem

pera

ture

(°C

)

qc"Top of TEC temperatureBottom of TEC temperature

Figure 2.18 Cooling flux produced by TEC during cooling from 94 to 54 °C, 13.8 V

input, cooling fan off

The maximum cooling flux for this period was 1.7 W/cm2. The cooling flux in

the case with the cooling fan behaves in a similar manner to the case with cooling fan on.

The maximum cooling flux is 0.4 W/cm2 lower in the case with no cooling fan, however.

29

The reason for this is because the heat sink is initially 18 K higher than the cooling fan

case meaning that ΔT is greater for the no cooling fan case.

2.2.3 Power consumption

The voltage and current consumed by the TEC were monitored via the display on

the power supply and a multimeter placed in series with the TEC.

The test used to determine the power and average power consisted of the TEC

mounted on the heat sink with the cooling fan on in one case and off in the other case.

Thermal grease was used between the TEC and heat sink. Control was provided by a

thermocouple placed on top of the TEC. The input voltage to the TEC for this test was

13.8 V. As the TEC was heated, the current was monitored on the power supply display.

The peak power for the TEC with and without cooling fan on was 23.7 W. The

average power over cycle for the TEC with cooling fan was 11.3 W. The average power

over cycle for the TEC without cooling fan was 7.9 W.

2.3 Independent TEC testing discussion

The results of the testing of the TEC are summarized in Table 2.2.

Table 2.2 Comparison of TEC with and without cooling fan

Max Heating

rate (K/s)

Average heating

rate (K/s)

Max Cooling

rate (K/s)

Average cooling

rate (K/s)

Max heating

flux (W/cm^2)

Max cooling

flux (W/cm^2)

Max Power consumption

(W)

Average power

consumption over cycle

(W)TEC with cooling fan 5.2 1.8 6.6 5.8 1.2 2.1 23.7 11.3TEC without cooling fan 5.8 2.0 5.2 4.3 1.9 1.7 23.7 7.9

30

The cooling fan produced a difference in the heating and cooling rates of the

TEC. The maximum and average heating rate were greater for the case without the

cooling fan. The reason for this is that the difference between the hot and cold side

temperatures of the TEC was less for the case without the heat sink fan on. This allowed

the top of the TEC to heat up more rapidly. The average and maximum cooling rates for

the TEC with cooling fan on were 1.5 K greater. This occurred because the difference

between hot and cold side temperatures was less for the case with cooling fan on.

The differences in heating and cooling rates translate into time for each cycle of

PCR. Using the TEC without the cooling fan would save 2.2 seconds in heating time for

each PCR cycle. Multiplying this value by 35 gives a total time saving of 77 seconds.

Using the TEC with cooling fan would save 2.5 seconds for each PCR cycle.

Multiplying this by 35 cycles gives a total time saving of 88 seconds for an entire PCR

run with 35 cycles. There is little benefit in using the cooling fan considering the time

savings.

The real savings in not using the cooling fan comes in power consumption. The

average power over cycle for the case without cooling fan was 3 W less than the case

with cooling fan on. The reason for this is because the average temperature of the TEC

remained warmer during operation without cooling fan. This made it possible to heat the

top surface of the TEC using less power. More power was required for cooling, but

overall the case with no cooling fan used less power.

A disadvantage of not using the fan on the heat sink was that the heat sink got

progressively warmer with each PCR cycle. This made the system slightly more difficult

to control.

31

The cooling fan on the heat sink produced an effect on the performance of the

TEC at steady state. The bottom of the TEC remained 29.4 K cooler in the case with the

heat sink fan on. The temperature differential between the hot and cold side of the TEC

with cooling fan was 17 K more than the TEC with no cooling fan. These results indicate

that the cool side of the TEC will rise above ambient during steady state operation. The

cooling fan keeps the bottom of the TEC cooler because it increases the convection

coefficient between the heat sink and the air.

2.4 Testing of a commercially available Peltier thermocycler, apparatus and

procedures

A commercially available thermocycler was tested to determine the characteristics

of the device. The thermocycler was a Techne Techgene thermocycler, model

FTGene5D rated at 120 or 230 V and 250 W shown in Figure 2.19.

Figure 2.19 Techne thermocycler analyzed in this section

32

The thermocycler was cooled and heated via 4 Marlow Industries Inc. TECs, part

number SP1975. On the bottom surface of the TECs was a 10.1 X 9.6 cm aluminum heat

sink with a fin height of 4.5 cm. On the top surface of the TEC an aluminum heating

block weighing 40.7 g was mounted. The dimensions of the heating block were 5.7 x 5.4

x 0.3 cm. The heating block had wells protruding from the surface. The wells were 11.9

mm high, had a diameter of 7.6 mm, and had a wall thickness of 0.8 mm. The heating

block contained 20 wells making it possible to do 20 different PCR reactions in one run

as shown if Figure 2.20.

Figure 2.20 Aluminum heating block mounted on top of TECs in Techne thermocycler

The thermocycler was tested using T type thermocouples. 10 µL of water was

inserted into a 200 µL tube. A thermocouple was then inserted into the solution and the

lid of the PCR tube was closed. This assembly was then inserted into one of the wells

located in the heating block. The thermocouple inside the tube measured the response of

33

the reaction vessel to the heating of the block. Another empty 200 µL tube was used to

measure the temperature of the heating block. A thermocouple was wedged between the

empty tube and the well in order to determine the temperature of the block. The lid of the

thermocycler was closed during the experiment.

The power to the device was measured by severing the power cable and inserting

a multimeter in series with the thermocycler to measure the current. The input voltage to

the device was 120 VAC.

2.4.1 Commercial Peltier thermocycler testing results

The commercial Peltier thermocycler was tested for heating and cooling rate. The

results of that test are shown in Figure 2.21.

Figure 2.21 Simulated PCR cycle using commercial Peltier thermocycler

45

50

55

60

65

70

75

80

85

90

95

100

225 240 255 270 285 300 315 330 345 360 375 390 405 420 435 450 465Time (seconds)

Tem

pera

ture

(°C

)

Inside 200µL tube 10 µLwaterBlock temperature,measured in well

34

The setpoint temperatures for this test were 94, 54, and 72 °C. The temperatures that

were measured via the thermocouples were all 2 K higher than these values. The error in

measurement is likely in the thermal cycler. The thermocouple is inserted into a hole in

the aluminum block and then glued into place. There may have been some air gaps or

other resistance present in this setup. The maximum heating rate within the well of the

heating block for the commercial device was 3.6 K/s. The maximum cooling rate turned

out to be 2.0 K/s. The average heating rate for the block of this device during heating

between setpoints 72 and 94 °C was 2.8 K/s and 1.8 K/s for heating between setpoints 54

and 72 °C/s. The average cooling rate for this device was 1.4 K/s.

2.4.2 Power consumption of the commercial device

The commercial Peltier unit was tested to determine the power requirements of

the device. The unit operated on 120 VAC. The current consumed by the device varied

during the different stages of operation. The current supplied to the device during

heating from 72 to 94 °C varied from 2.2 to 1.2 A with the largest current consumption at

the beginning of heating and the smallest current as the unit reached the setpoint. This

makes sense because it is expected that the control system of the device would scale the

power down as the setpoint temperature was reached. Current during heating between 54

and 72 °C peaked at 2.0 A during the initial phase of heating and decreased to 0.3 A as

the device neared the setpoint. During cooling, the current in the device ranged from 1.8

A to1.1 A as the setpoint was reached. The current consumed by the device at the 94 °C

plateau was 0.4 A, 0.3 A at the 72 °C setpoint, and 0.3 A at the 54 °C setpoint.

35

Considering these values gives a maximum power requirement of 264 W with an average

power over cycle of 145.0 W.

2.4.3 PCR results with the commercial device

PCR was run in the commercial device many times. The results presented in this

section are just a representative sample of those efforts. An experiment was run that

demonstrated the effect of elongation time on PCR results. As was discussed in Section

1.2, two different sets of primers were used. One primer set was for amplifying a piece

of DNA 879 bp long while the other primer set was for a 216 bp amplicon. The

elongation time was adjusted until the smaller section of DNA was amplified, but the

longer section was not. The protocol for these experiments is shown in Table 2.4.

Table 2.4 Protocol for elongation time experiment

879 bp amplicon reaction mixture (µL)

216 bp amplicon reaction mixture (µL)

Buffer 3.0 3.0MgCl2 1.0 1.0dNTP 4.5 4.5Primers 1.5 1.0DNA 1.0 1.0Taq Polymerase 0.3 0.3Albumine 3.0 3.0Water 10.7 11.2

Total 25.0 25.0

PCR was run in either 500 or 200 µL polypropylene PCR tubes with the heated lid on the

Techne thermocycler on to prevent solution evaporation. The PCR cycle for the first case

consisted of a 95 °C activation period for 15 minutes followed by 35 cycles of 30 seconds

at 94 °C, 30 seconds at 54 °C, and 40 seconds at 72 °C. A 15 minute final elongation

step at 72 °C completed the PCR run. The Techne control system only counts the plateau

36

time in the cycle time. For example, 30 seconds at 72 °C means that the thermocycler

plateaued at that temperature for 30 seconds. The time required to heat the block from 54

to 72 °C is not included. The results of this test are shown in Figure 2.22.

Lane Description0 DNA ladder1 500 µL tube, 10 µL solution, 879 bp amplicon2 500 µL tube, 10 µL solution, 216 bp amplicon3 200 µL tube, 5 µL solution, 879 bp amplicon4 200 µL tube, 5 µL solution, 216 bp amplicon5 200 µL tube, 5 µL of solution, 879 bp amplicon6 200 µL tube, 5 µL of solution, 216 bp amplicon

Figure 2.22 PCR reaction results for extension time test, 40 s elongation time

The results indicate that 40 seconds at 72 °C was sufficient to elongate both the 216 and

879 bp DNA fragment. The DNA ladder may be used to determine the length of the PCR

amplicon. The DNA ladder is a 1000 bp ladder in 100 bp increments. This means that

the line in lane 0 closest to the bottom of the picture is 25 bp, the next line is 100 bp, the

next line is 200 bp, the next line is 300 bp, all the way up to the first line that is the 1000

bp marker.

25 bp

100 bp

200 bp

1000 bp

37

The next test run in this sequence of tests to determine the amplification rate of

these two amplicons was to decrease the elongation time to 30 seconds, decrease the

denaturation time to 20 seconds, and decrease the annealing time to 20 seconds. The

results of that test are shown in Figure 2.23.



The results indicate that decreasing the elongation time to 30 seconds has no effect on

either amplicon. It is expected that as the elongation time continues to be decreased, the

longer amplicon will fail to be copied. Decreasing the denaturation and elongation times

from 30 to 20 seconds also had no effect on the reaction as a signal was received for each

tube.

25 bp

100 bp

200 bp

1000 bp

38

The third step in this test consisted of decreasing the denaturation and annealing

times to 15 seconds and decreasing the elongation time to 25 seconds. The results of that

test are shown in Figure 2.24.



The lack of a clear signal for any of the 879 bp amplicons in Figure 2.24 indicates that

the 25 second elongation time was a problem for the longer amplicon. Theoretically, the

copying rate for DNA polymerase is 50 bp/s, which would mean that it would be possible

to copy a 1250 bp segment in this time. The experiment that was run in this case

indicates that the rate of copy is around 29 bp/s.

An experiment was run to determine if the results shown in Figure 2.24 were due

to the decrease in annealing and denaturation times, or by the decrease in extension time.

25 bp

100 bp

200 bp

1000 bp

39

The denaturation and annealing time in this step remained at 15 seconds while the

elongation time was increased to 30 seconds, 5 seconds greater than the previous test.

The results of the test are shown in Figure 2.25.


Figure 2.25 PCR reaction results for extension time test, 30 s elongation time with 15 s

at denaturation and annealing steps

The results indicate that with an elongation time 5 s longer, the longer amplicon was

copied, even with the shorter denaturation and annealing steps. This result shows that the

limiting factor in PCR is the elongation phase. The copying speed of DNA polymerase is

a limiting factor in PCR speed.

Another method of visualizing PCR reaction success is by running the reaction

with ethidium bromide in the solution at a volumetric concentration of 1:4000 in a glass

25 bp

100 bp

200 bp

1000 bp

40

capillary. This eliminates the need to use gel electrophoresis to determine if the reaction

occurred. If the reaction occurred, the solution will fluoresce when placed in a UV light

enclosure. If it did not occur, then no fluorescence will be seen. Ethidium bromide is a

substance that fluoresces when exposed to UV light. Ethidium bromide has the ability to

attach itself to the DNA molecule (D'Amico, Paiotta et al. 2002). The ethidium bromide

undergoes an increase in flourescense when bound to the DNA molecule (Pyun and Park

1985). The concentration of ethidium bromide must be just right. If a large

concentration of ethidium bromide is used, then it will be impossible to distinguish

between a successful reaction and an unsuccessful reaction because ethidium bromide

fluoresces under UV light by itself. If too small an amount is used then no fluorescence

will be seen if the reaction occurred or if if did not. A picture of three capillaries in the

UV light hood is shown in Figure 2.26.

Figure 2.26 Glass capillaries run with a 1/4000 volumetric concentration of ethidium

bromide. 1) PCR solution before amplification, 2) PCR solution after amplification in

commercial thermocycler, 3,4) Unsuccessful reactions on microheater

41

The fluorescence is easily detectable in the successful PCR reaction, labeled 2 in Figure

2.26. There is no fluorescence in the reaction mixture before PCR, labeled 1, and the two

unsuccessful reactions run on the microheater, labeled 3 and 4 in Figure 2.26.

42

3. TESTING OF A CONVECTIVE THERMOCYCLER FOR PCR

A convective thermocycler was built in the lab to test the possibility of using a

device that directs an air flow past a sample to perform the heating and cooling for the

PCR reaction. The heating and cooling rate, heating and cooling flux, and power

consumption were monitored.

3.1 Testing of a convective thermocycler, apparatus and procedures

Two different configurations of the convective thermocycler were built. The first

variation is shown in Figures 3.1 and 3.2.

Figure 3.1 Convective thermocycler setup #1

17.1 cm OD AC fan

10.2 cm OD X 58.4 cm aluminum vent pipe

Copper-constantan type T thermocouple


Omega engineering solid state relay part No. SSR240DC215

Transformer Type 3PN1010

43

Figure 3.2 Heater mounted within vent pipe

The second configuration was designed with two fans. One of the fans was used

for flowing air over the heating element while the other fan was used for cooling. The

purpose of this was to speed cooling. This double fan convective thermocycler is shown

in Figures 3.3 and 3.4.

Figure 3.3 Double fan convective thermocycler setup

Heating element removed from hair dryer

10.2 cm OD x 55.2 cm aluminum vent pipe

17.1 cm OD AC fan

Copper-constantan type T thermocouple


Omega engineering solid state relay part No. SSR240DC25

Transformer Type 3PN1010

44

Figure 3.4 Heater element within double fan convective thermocycler

There are several similarities between the two convective thermocycler designs.

The convective thermocyclers described in this section were controlled with pulse width

modulation (PWM), similar to the thermoelectric cooler (TEC) described in Chapter 2. A

Visual Basic program was written to control the system. The software used a

proportional-integral-derivative (PID) controller scheme to control the fractional on-time

of the heater during each control period (1 sec control period used in this work). The

pipe centerline air temperature near the outlet was measured and was compared against

the setpoint to act as the controller input signal.

The fan on the single fan convective thermocycler was used to drive air through

the device. The air is heated as it flows past the heating element. The fan speed in the

single fan unit was also controlled with PWM in heating mode. The fan power was

turned on only 50% of the time to yield a reduced fan speed. During cooling, the heater

Ambient air channel

Thermocouple for temperature control

Heating element from hair dryer

Heated air channel

45

was turned off and the fan was run at full power to introduce the maximum flow of

cooling air.

The double fan convective thermocycler operated in much the same way as the

single fan convective thermocycler. The heater was controlled in an identical manner.

The difference between the two designs was that the double fan convective thermocycler

used two fans. The first fan was run at reduced speed as described above for the single

fan unit and was responsible for driving ambient air over the heating element to create a

stream of hot air in the main channel (second fan was off during heating). The second

fan was responsible for providing the cooling air flow. It was run at full speed during

cooling mode (the heating fan was turned off during cooling).

The software recorded data at a frequency of 1 Hz. The data recorded was the

pulse width, setpoint, and temperature at the outlet of the heater. An Extech Instruments

multimeter, model 22-816 was used to measure the voltage and current output of the

variable transformer.

The thermocouple for measuring the temperature of the air in the thermocycler

was placed along the axis of the pipe at a point 7.6 cm upstream from the end of the pipe.

3.2 Convective thermocycler test results

The maximum heating rate for the single fan convective thermocycler was 2.0

K/s. The maximum cooling rate was 1.8 K/s. A single simulated PCR cycle using the

single fan convective thermocycler is shown in Figure 3.5.

46

45

50

55

60

65

70

75

80

85

90

95

100

1325 1340 1355 1370 1385 1400 1415 1430 1445 1460 1475 1490 1505 1520 1535 1550

Time (Seconds)

Tem

pera

ture

(°C

)

Air TemperatureSetpoint

Figure 3.5 Simulated PCR cycle using the single fan convective thermocycler

The average heating and cooling rates were also calculated. The average heating

rate for heating between 54 and 72 °C for this setup was 1.2 K/s. The average heating

rate between 72 and 94 °C was 0.9 K/s. The average cooling rate for the single fan

convective thermocycler was 1.2 K/s.

The double fan convective thermocycler was tested to determine the heating and

cooling rates. The results of the test are shown in Figure 3.6.

Figure 3.6 Simulated PCR cycle using the double fan convective thermocycler

47

Vent pipe Mixing walls

Fan

Heater

Figure 3.7 Cross section of

single fan convective

thermocycler showing mixing

walls located within vent pipe

The maximum heating rate of the double fan convective thermocycler was 4.2 K/s. The

maximum cooling rate for this setup was 7.9 K/s. The average heating rates were 2.6 K/s

for heating between 54 and 72 °C and 2.0 K/s for heating between 72 and 94 °C. The

average cooling rate was 5.3 K/s. The heating rates are higher because the relative flow

rate and power input were set to obtain higher rates for the double fan unit. Cooling rates

are higher because the cooling fan flow rate is higher due to a more open flow path

obtained when the heater element is not present in the cooling flow path. These

characteristics were the motivation in designing the double fan unit.

One of the important characteristics of a convective thermocycler is temperature

uniformity across the area where the reaction will be occurring. The temperature should

be as uniform as possible to ensure a quality PCR reaction. The single fan convective

thermocycler was tested for temperature uniformity. Different ideas were tried in order

to make the radial temperature profile as even as possible. 2 in. thick foil and fiberglass

duct insulation was used in one test to cover the outside of the duct. This was done in

order to reduce the amount of heat loss

through the walls of the duct, thus

making the wall temperature closer

to the duct midpoint temperature.

Another variant of the convective thermocycler

was to place mixing walls within the duct in an

attempt to make the flow within the tube more

turbulent, thereby enhancing mixing as shown in

Figure 3.7.

48

Another tactic that was tested to make the temperature profile more uniform was to add

an additional vent pipe 61 cm long to the end of the convective thermocycler. The result

of the testing is shown in Figure 3.8.

Figure 3.8 Radial temperature measurements for single fan convective thermocycler

with heater and fan at duty cyle = 1

The configuration that provided the best results in terms of temperature uniformity was

the case where the pipe was covered with insulation, mixing walls were inserted into the

pipe, and no extension was used. The standard deviation for this case was 3.5 K,

compared to 3.6 K for the case with insulation, no mixing walls, and no extension. The

worst case scenario was the case with the extension, no insulation, and no walls. This

case had a standard deviation of 9.3 K. The effect of the mixing walls on the temperature

49

uniformity was very small, as can be seen by comparing the standard deviations for the

top two traces in Figure 3.8. Only 0.1 K separated the standard deviations of these two

cases. There was a large drop off in the standard deviation in temperature profile for the

case with mixing walls, no extension or insulation. The standard deviation for that case

was 5.1 K, indicating that the insulation was more important than mixing walls in

maintaining temperature uniformity.

3.2.1 Power consumption of convective thermocyclers

For the single fan unit, the power input to the fan was 108 VAC and 0.6 A. The

power supplied to the heater element was 78.4 VAC at 7.7 A. For the double fan

convective thermocycler, the heater was supplied with of 118.2 VAC which drew a

current of 11.6 A. The cooling and heating fans were supplied with 82.3 VAC each of

which drew a current of 0.4 A.

The power consumption of the single fan convection cycle was found by

multiplying the current, voltage and duty cycle. The power required to run the fan was

not taken into account for this calculation because the power required by the fan is small

compared to the power required by the heater. The power required to run the fan at full

power is only 10% of the power required to run the heater at full power for the single fan

convective thermocycler. The single fan convective thermocycler heater required 604 W

to operate when it was not pulsed. The average power over a cycle shown in Figure 3.5

was 255 W.

The power consumption for the double fan convective thermocycler was found

similarly. The power required for the heating and cooling fans was not taken into

50

account for this calculation. The power required to run the fan at full power is only 2.4%

of the total power required to run the heater of the double fan convective thermocycler at

full power. The double fan convective thermocycler required 1371 W of power when not

pulsed. The average power over cycle shown in Figure 3.6 was 645 W.

It is noted that the power requirements of these convective thermocylers is quite

high due to the fact that a large volumetric flow rate of heated air is dumped to the room.

Air heating/cooling requires significant air velocities to maintain high heat transfer

coefficients and such designs suffer from poor energy performance (high power

requirements) unless some recycling scheme is used to recycle the hot air instead of

throwing it away. In early designs, energy recycling was attempted but it was found that

such designs impacted the cycling speed. Air heating/cooling designs were found to

provide reasonably fast heating and cooing but at the cost of large power requirements.

3.2.2 Convection characteristics of the thermocycler

The velocity of the air within the two convective thermocyclers was found in

order to determine the heating power produced at the tube outlet. A control volume

analysis was used to determine the velocity of the air in the tube. A schematic of the

model used is shown in Figure 3.9.

Figure 3.9 Schematic of control volume used for air velocity determination

Fan

Heater q

AV Furnace duct

Control volume

Outlet

Inlet

51

where VA is the air velocity produced by the fan and q is the energy per unit time

transferred to the control volume by the heater. The control volume energy rate balance

reduces to (Moran and Shapiro 1999)

)(0.

oiCV hhmq −+−= (3.1)

assuming that the control volume is at steady state, ignoring kinetic and potential energy

effects, and ignoring heat transfer from the control volume. Here .

m is the mass flow rate

and is given by (Moran and Shapiro 1999)

νaveAV

m =.

(3.2)

where ih and oh are the enthalpies of the air at the inlet and outlet respectively. A is the

cross sectional area of the pipe, Vave is the average air velocity within the pipe, and ν is

the specific volume of the air in the mass flow rate equation. The specific volume of the

air may be determined by considering the air to be an ideal gas at atmospheric pressure.

The enthalpies were also found considering air to be an ideal gas.

Each convective thermocycler was run at steady state to determine the air velocity

during heating within the thermocycler tube. For these tests, the heaters on both the

single and double fan convective thermocycler were run at constant power (25% on

time). The heating fan on both units were run at the same settings that were used in the

thermal testing (reduced speed at 50 % power).

A test was run with both thermocyclers to determine the air velocity in cooling

mode. The fan on the single fan convective unit was run at full power with the heater run

at a 50 % duty cycle to determine the air velocity in cooling mode. The cooling fan of

the double fan convective thermocycler was also run at a full power during cooling. This

52

cooling fan did not have a heater obstructing its path, so the air velocity was determined

with the manufacturing specifications. The manufacturing specifications indicate that the

fan is capable of driving 235 cfm of air at 0 static pressure (Rotron 2005). This value

was used by considering an analysis of the flow in the vent tube. The pressure drop in

the vent tube may be found using the Moody friction factor equation (Incropera and

DeWitt 1996). The equation is

2/)/(

2aveV

DdxdPfρ

−= (3.3)

Where P is the pressure, x is the distance along the vent tube, D is the diameter of the

tube, ρ is the density of air taken at ambient, Vave is the average air velocity, and f is the

friction factor. The friction factor may be obtained from the Moody diagram and is

dependent on the Reynolds number. Some assumptions were made to get an idea of the

pressure drop in the tube. For an air velocity of 4 m/s, which is an overestimate

considering the specifications and the fan diameter, the Reynold’s number is 25,000.

This leads to a value for the pressure drop of 2.19 Pa using equation 3.3 and knowing the

diameter and length of the vent tube. The airflow is near the maximum value for the fan

considering the manufacturers static pressure versus airflow curve (Rotron 2005).

Therefore, the air velocity was found from the maximum volumetric flow rate by

dividing by the area of the fan.

The temperature at the outlet of the heating tube during steady state was recorded

in order to determine the enthalpy of the air at the exit during all air velocity

determinations except for the double fan convective thermocycler cooling fan. The inlet

temperature of the air was room temperature. This allowed the enthalpy to be set at the

inlet and the outlet. The control volume heat transfer was simply the power supplied to

53

the heater. The only unknown variable in Equation 3.1 was the air velocity. The results

of the air velocity analysis are shown in Table 3.1

Table 3.1 Results of air velocity analysis for single and double fan convective

thermocyclers

Single fan heater

Single fan cooler

Double fan heater

Double fan cooler

Air velocity (m/s) 0.5 0.7 1.2 3.4Fan input power (W) 32.4 64.8 16.5 32.9

The air velocities in Table 3.1 are also given with the fan input power in order to

demonstrate the relationship between the two. The fan input power was found by

multiplying the input voltage, current, and duty cycle of the fan. The air velocity of the

single fan cooler is lower than the double fan heater air velocity. The single fan cooler

fan was run at full speed while the double fan heater was run at 50 % duty cycle. The

flow path in the single fan convective thermocycler was more obstructed by the heater

than the double fan convective thermocycler because of the method of attachment. The

double fan heater also used 4 times less power than the single fan cooler. This

demonstrates that it is important to avoid obstructing the heating fan too much with the

heating coil. The double fan cooler supplied an air velocity that was nearly three times

greater than the double fan heater using only twice as much power. This was because the

double fan cooler had no obstruction in its path.

The next variable found to characterize the convection thermocycler system was

the Nusselt number. The reaction vessels that were placed inside the convection

thermocycler were cylindrical or cylinder like, therefore a cylindrical Nusselt correlation

was used. The correlation was (Incropera and DeWitt 1996)

54

4/1____

PrPrPrRe ⎟⎟

⎠

⎞⎜⎜⎝

⎛=

s

nmDD CNu (3.4)

where ____

DNu is the average Nusselt number over the surface of the cylindrical reaction

vessel, Pr is the Prandtl number, Prs is the Prandtl number evaluated at the surface

temperature of the vessel, and C, m, and n are constants based on ReD and Pr. The

temperature of the surface of the reaction vessel was unknown. This fact does not play

heavily in the analysis, however. The only parameter evaluated at the surface

temperature in Equation 3.4 is PrS. The largest temperature difference that could possibly

be encountered in this analysis is 40 K, the difference between the denaturation

temperature of 94 °C and the annealing temperature of 54 °C. The Prandtl number varies

by only 0.006 between these temperatures, therefore the surface temperature of the vessel

was assumed to be the setpoint temperature prior to the one being analyzed. For

example, if the air temperature within the tube was at the 72 °C setpoint, then the surface

temperature of the vessel was assumed to be 54 °C. All of the other parameters in the

equation were evaluated at the air temperature within the vent tube. The diameter of the

reaction tube in this case was taken to be 2.6 mm, which is the average diameter of the

bottom 2.5 mm of a 200 µL polypropylene PCR tube. The bottom of the tube was

measured only because this was the area that was filled when 10 µL of water were

inserted into it as shown in Figure 3.10.

55

Figure 3.10 200 µL PCR tube with 10 µL of water used in convective analysis. All

dimensions in mm

The Nusselt number and the average convection coefficient were found for the air

velocities shown in Table 3.1. The results are summarized in Table 3.2.

Table 3.2 Nusselt number and convection coefficients for single and double fan

convective thermocycler

3.2.3 System response with a test sample

The lumped capacitance method was used to derive an expression for the time

constant of a contained water sample in the convective thermocycler that is useful for

discussing the system dynamics. The lumped capacitance approximation is valid if the

Biot number is less than 0.1 (Incropera and DeWitt 1996). In this case, the Biot number

56

was evaluated for a cylindrical reaction vessel with an average diameter of the bottom 2.5

mm of a 200 µL polypropylene PCR tube. The polypropylene tube is 0.3 mm thick. The

thermal conductivity of this system was lumped together based on the volumetric

percentage of water and polypropylene. This gives a thermal conductivity of

0.47 W/m K, considering that the thermal conductivity of water is 0.60 W/m K and

polypropylene is 0.17 W/m K. Evaluating the Biot number with these constraints gives a

value of 0.17 with an average convection coefficient of 120 W/m2K and 0.1 with an

average convection coefficient of 71 W/m2K. These results indicate that the lumped

capacitance model may be a poor choice for the cooling case in the double fan convective

thermocycler, but may be a good fit for the other situations. Applying an energy balance

on the reaction tube and sample, the following equations can be derived (Incropera and

DeWitt 1996)

⎥⎦

⎤⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−=

−−

∞

∞ tVc

hATTTT s

i ρexp (3.5)

where T is the temperature at time t, Ti is the initial temperature of the system, ∞T is the

air temperature within the vent pipe, h is the convection coefficient, ρ is the density of the

reaction vessel, V is the volume of the reaction vessel, c is the specific heat of the

reaction vessel, and As is the surface area of the reaction vessel. This expression gives

rise to an expression for the time constant of the reaction vessel

ttt CR=τ (3.6)

The equation states that the time constant of the system is equal to the resistance to

convection multiplied by the total thermal capacitance of the system. The analysis that

follows will be performed first on a 200 µL polypropylene PCR tube with 10 µL of water

57

inside of it modeled as a cylinder of diameter 2.6 mm and height 2.5 mm. The time

constant in this case is given by

s

ppppwpwwt Ah

fcfcV )( ρρτ

+= (3.6)

where V is volume of the reaction tube, ρw and ρp are the densities of the water and

polypropylene, cpw and cpp are the specific heats of the water and polypropylene, As is the

surface area of the cylinder, and fw and fp are the fraction volume of water and

polypropylene in the reaction volume. A lumped thermal conductivity was found using

the volumetric fraction of water and polypropylene that was used to determine the

average heat transfer coefficient from DNu .

ppww fkfkk += (3.7)

Time constants were found for the system using the average convection

coefficients found in Table 3.2. The results are shown in Table 3.3.

Table 3.3 Calculated time constants (τ) for the air velocities within the single and double

fan convective thermocycler for a modeled 200 µL polypropylene tube with 10 µL of

water

The time constant is the time that it takes the reaction volume to reach within 63.2 % of

the difference between the initial temperature of the solution and the setpoint

temperature. If the setpoint was 72 °C, the initial temperature was 52 °C, and the time

constant was 34 seconds, the reaction should be at 64.6 °C after 34 seconds, for example.

58

A 200 µL polypropylene PCR tube filled with 10 µL of water was modeled using

a 2-D heat transfer finite element program. A schematic of the model used is shown in

Figure 3.11.

Figure 3.11 2-D model used for convective analysis of 200 µL polypropylene tube filled

with 10 µL of water, dimensions in mm

The temperature profiles of the convective thermocyclers obtained in the results section

were input into the 2-D model shown in Figure 3.11 along with the time average heat

transfer coefficients, calculated with the velocity and temperature information from

experiment. The experimental results were then compared with the 2-D model results to

determine the validity of this method. The properties of polypropylene used in the model

are shown in Table 3.4 (Goodfellow 2005).

Table 3.4 Polypropylene properties

Density (kg/m3)

Thermal conductivity

(W/m K)

Specific heat (J/K kg)

Polypropylene 900 0.17 1800

59

Heating from 52 to 94 °C was modeled with the 2-D finite element program for

the single fan convective thermocycler. The experiment with a 200 µL tube with 10 µL

of fluid in the single fan convective thermocycler indicated that the initial temperature of

the fluid before the heating began was 53.3 °C as shown in Figure 3.12.

Figure 3.12 Experimental results for single fan convective thermocycler with 200 µL

tube filled with 10 µL of water placed inside vent pipe

The temperature of the fluid inside the tube was measured with a thermocouple placed

inside the tube, immersed in the fluid. Analysis showed that the average convection

coefficient over the surface of the cylinder remained relatively constant during heating

from 52 to 94 °C, varying by only 0.06 %. A constant value of h , average convection

coefficient, was used for this reason and was taken to be 48.0 W/m2K.

The 2-D model produced the results shown in Figure 3.13.

60

Figure 3.13 2-D model correlation with experimental results for single fan convective

thermocycler with 200 µL PCR tube containing 10 µL of water during heating from 52 –

94 °C

The maximum error between the experimental value and the modeled value was 3.6 K.

The calculated time constant for the system was 34 seconds using the lumped capacitance

model. This means that the temperature at 34 seconds should be 64.6 °C for the heating

case between 52 and 72 °C. The actual water temperature was 65.2 °C at this point while

the 2-D finite element model predicted a temperature of 65.7 °C. The significant time lag

of the reaction mixture in comparison with the air temperature is shown in Figure 3.13.

The air temperature reached the setpoint of 72 °C within 8 seconds while the reaction

mixture failed to reach the setpoint temperature within the 60 second time frame.

The single fan convective thermocycler with 200 µL PCR tube filled with 10 µL

of water during cooling from 86 to 52 °C was modeled with the 2-D finite element

61

modeling package. The initial temperature of the solution at the beginning of the

temperature decrease was 86.2 °C. The convection coefficients used were 57 W/m2K for

the first fifteen seconds of the cool down, corresponding to the time when the fan was on

at full power, and 48 W/m2K for the period after 15 seconds when the fan was pulsed.

The results of the analysis are shown in Figure 3.14.

Figure 3.14 2-D model correlation with experimental results for single fan convective

thermocycler with 200 µL PCR tube during cooling from 86 – 52 °C

The calculated time constant for this reaction volume during cooling was 29

seconds, indicating that the solution temperature should be at 64.5 °C after one time

constant. The temperature of the solution was actually at 62.1 °C at this time. The 2-D

modeled temperature was at 71 °C after 29 seconds. The maximum error between the

model and the actual temperatures was 8.8 K, indicating poor agreement between the

62

model and experimental data. This result indicates that the actual air velocity in the tube

may be higher than calculated using the control volume approach.

The double fan convective thermocycler reaction volume was modeled with the

lumped capacitance method in order to validate this method for future work. The

performance of the double fan convective thermocycler with a 200 µL polypropylene

tube filled with 10 µL of water is shown in Figure 3.15.

Figure 3.15 Experimental results for double fan convective thermocycler with 200 µL

polypropylene tube filled with 10 µL of water placed inside vent pipe

Recall that the polypropylene tube was modeled as a cylinder with a diameter equal to the

average 2.5 mm of the reaction tube and a wall thickness of 0.3 mm. The time constant

for the air in the vent tube was 8 seconds for the heating modes and 6 seconds for the

cooling mode. The calculated time constants for the fluid in the reaction tube are 23

63

seconds for the heating mode and 14 seconds for the cooling mode when the fan is on at

full power as shown in Table 3.3.

The temperature at time t is dependent on ∞T , the temperature of the air stream, as

shown in Equation 3.5. The temperature of the air stream is transient during the initial

part of the heating phase. This gives two views on how to model this problem. The first

view is to model the problem with the transient air temperature of the air, substituting the

appropriate value of ∞T (t). The second view is to model the problem using constant

value representing the setpoint for ∞T since the time constant of the reaction volume is 3

times greater than the time constant of the air within the vent tube during heating. These

two approaches were used for the heating modes of the double fan convective

thermocycler. The results of the modeling are shown in Figure 3.16.

45

50

55

60

65

70

75

80

85

90

95

100

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150Time (seconds)

Tem

pera

ture

(°C

)

Air temperature

Water temperature

Lumped capacitancemodel, constant Tinf

Lumped capacitancemodel, transient Tinf

Figure 3.16 Lumped capacitance model with experimental results for double fan

convective thermocycler with 200 µL PCR tube during heating from 54–94 °C

64

The correlation between experimental data and the lumped capacitance models is

strong. The lumped capacitance model with constant ∞T has more error associated with

it then the model with transient ∞T during the first eight seconds of the test, and again

during the first eight seconds of heating from 72 to 94 °C. The average error over the

first 8 seconds of heating from 54 to 72 °C for the constant ∞T case was 1.0 °C while the

average error associated with the transient ∞T case was 0.4 °C. The average error over

the first 8 seconds of heating from 72 to 94 °C for the constant ∞T case was 2.0 °C while

the average error for the transient ∞T case was 0.5 °C. The average error for the

remainder of the time period was also calculated. The average error for the transient ∞T

case turned out to be 0.5 °C while the average error for the constant ∞T case turned out to

be 0.6 °C. The time constant of the reaction volume for heating between 54 and 72 °C

was 20 seconds. The time constant of the reaction volume for heating between 72 and 94

°C was 25 seconds. The calculated time constant was 23 seconds for these temperature

intervals. These results indicate that accurate results can be obtained by ignoring the

transient heating of the air in modeling the solution temperature with the lumped

capacitance method as long as the time frame being analyzed is outside the time constant

of the air, in this case 8 seconds, and the time constant of the reaction volume is 3 times

larger than the time constant of the air.

The lumped capacitance model was used to model cooling of the reaction volume

consisting of a 200 µL PCR tube containing 10 µL of water within the double fan

convective thermocycler. The same approach was used as was used for the heating

modes of the double fan convective thermocycler. The first situation was when ∞T was

65

held constant at the setpoint temperature. The second situation was when ∞T was allowed

to vary with the air stream temperature. The cooling phase consisted of two different

values of h . The first value was the one associated with the fan at full power and was

taken to be 120 W/m2 K. Recall that during cooling, the cooling fan is turned on at full

power and the heater fan is turned off. The second value of h was associated with the

condition when the cooling fan was turned off and the thermocycler returned to its

normal operating state, the heating fan pulsed with the heater on. The second value of h

was taken to be 71 W/m2K. The cooling fan was on at full power during the test for

seven seconds. The results of the modeling using the lumped capacitance method are


45

50

55

60

65

70

75

80

85

90

95

100

0 15 30 45 60 75Time (seconds)

Tem

pera

ture

(°C

)

Air Temperature

Water temperature

Lumped capacitance model,constant TinfLumped capacitance model,transient Tinf

Figure 3.17 Lumped capacitance model with experimental results for double fan

convective thermocycler with 200 µL PCR tube during cooling from 94–54 °C

66

The time constant for the cooling air was 6 seconds because it took the air

temperature that long to reach 69 °C. The time constant for the reaction volume was 17

seconds. The calculated time constants for the fan at full power and the heating fan at 0.5

duty cycle were 14 and 23 seconds, or an average of 18 seconds, which corresponds well

with the experimental results. It is clear by looking at Figure 3.17 that the error

associated with the constant ∞T case was much greater than the error associated with the

transient ∞T case over the six second time constant of the air temperature. The average

error for the time period minus the initial six seconds for the transient ∞T case was 2.4 °C

while the average error for the constant ∞T case was 1.4 °C. This indicates, as does the

heating case, that the transient cooling of the air may be ignored when modeling the

reaction volume using the lumped capacitance method for times much greater than the

time constant of the air. If the reaction volume temperature is desired during a time less

than the time constant of the air, than the transient ∞T lumped capacitance model should

be used. Both lumped capacitance models in Figure 3.17 have a small area where the

curve is not smooth at t = 8 s. This is because this is the point where the heat transfer

coefficient was changed, corresponding to the time when the cooling fan was turned off,

and the heating fan was turned on, pulsed at 0.5 duty cycle.

3.3 Convective thermocycler options to enhance performance

Increasing the heating rate of the air within the vent tube is one way to enhance

the performance of the thermocycler. Considering the heater itself, a heater with greater

surface area could be designed to allow for more efficient heat transfer from the heater to

67

the air. More power could also be supplied to the heater to add more energy to the

system.

High air velocity and high heating rate of the air is a challenge because the heater

can only supply so much energy to the air as it flows by it. Two important parameters for

high reaction volume heating rate are velocity and air temperature heating rate. The

velocity of the air may only be increased by so much before the heater will not be able to

heat the air to the setpoint temperature.

An increase in air velocity within the convective thermocyclers would increase

the Reynolds number and would therefore increase the convection coefficient between

the tube and the air. The time constant of the reaction was determined for a variety of

different air velocities assuming a glass cylinder with wall thickness of 240 µm and

outside diameter of 890 µm corresponding to the diameter of a commonly found glass

capillary, an air stream temperature of 94 °C, a surface temperature of 72 °C, and a length

of 2.6 mm. The capillary was filled with water. The plot of the time constant versus air

velocity is shown in Figure 3.18.

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8 9 10Air velocity (m/s)

Tim

e co

nsta

nt (s

)

Figure 3.18 Plot of time constant versus air velocity for cylindrical configuration with

constant diameter, length, wall thickness, surface temperature, and air temperature

68

All of the time constants, except one, graphed above are lower than the heating time

constant of the convective thermocycler which was 8 seconds. The cooling time constant

of the double fan convective thermocycler as tested was 6 seconds. All of the graphed

time constants are less than this also, except for the 0.5 m/s case. The change in time

constant for this reaction vessel versus air velocity levels out somewhat after 4 m/s. The

benefit of increasing the air velocity after this point results in reduced return, compared

with the benefit obtained before the 4 m/s point.

The reaction volume may be manipulated to enhance the heat transfer. One way

to do this for a cylindrical reaction vessel is to decrease the diameter of the reaction

vessel. It was shown above that a cylindrical glass tube with an outside diameter of 890

µm has a shorter time constant compared with the 2.6 mm diameter polypropylene tube.

A plot of time constant versus cylinder diameter was made assuming a constant air

velocity of 1.2 m/s, a wall thickness of 244 µm, a polypropylene tube, and a length of 2.6

mm is shown in Figure 3.19.

0

50

100

150

200

250

300

350

0.0 5.0 10.0 15.0 20.0Cylinder diameter (mm)

Tim

e co

nsta

nt (s

)

Figure 3.19 Plot of time constant versus diameter for cylindrical polypropylene

configuration

69

This plot shows that the time constant with a cylinder of this diameter would be 5

seconds which is less than the double fan convection air temperature time constant of

eight seconds. The cylindrical tube studied in Secion 3.2.3 is the point with diameter =

2.62 mm corresponding to a time constant of 23 seconds. This analysis shows the

dramatic effect that changing the diameter of the reaction vessel can have on the time

constant of the system. The cost of decreasing the diameter of the reaction vessel is that

one can get less solution into it per unit length.

The lumped capacitance method was used to model the behavior of a glass

capillary in the double fan convective thermocycler. The inside diameter of a commonly

available glass capillary is 0.4 mm so it would be necessary to have a capillary 39.8 mm

long to have 5 µL of PCR solution in it. For a reaction volume of this geometry with this

amount of water in it, a wall thickness of 244 µm, and an air velocity of 1.2 m/s the

lumped capacitance method yields a time constant for the system of 4.8 seconds, a value

of 3.6 for the average Nusselt number, and 126 W/m2K for the average convection

coefficient. This time constant was used to model the behavior of the reaction volume in

the double fan convective thermocycler as tested previously. The results of the analysis

are shown in Figure 3.20.

70

Figure 3.20 Lumped capacitance model of reaction volume temperature for glass

capillary of 890 µm outside diameter with 5 µL of water inside

These results indicate that the solution temperature is nearly identical to the air

temperature for the case with the glass capillary in the double fan convection

thermocycler. This indicates that the limiting factor in PCR speed is the ability of the

thermocycler to heat the air since the time constant of the reaction mixture is less than

that of the air temperature within the thermocycler.

The reaction mixture may also be placed into a reaction vessel of different shape.

One reaction vessel that is of interest is a sphere. A sphere of inside diameter 2.12 mm is

needed to have a reaction volume of 5 µL. the average Nusselt number is given by

(Incropera and DeWitt 1996)

71

4/14.03/22/1 Pr)Re06.0Re4.0(2 ⎟⎟

⎠

⎞⎜⎜⎝

⎛++=

sDDDNu

μμ (3.8)

The lumped thermal conductivity was used in the calculation of the average heat transfer

coefficient as in Equation 3.7. A glass sphere with this inside diameter and wall

thickness of 244 µm, an air velocity of 1.2 m/s, a surface temperature of 72 °C, an air

stream temperature of 94 °C, and the lumped capacitance method, a time constant of 16.1

seconds is obtained with an average Nusselt number of 7.2, and an average value of the

convection coefficient of 90 W/m2K. The time constant for the spherical geometry is

greater than the 890 µm outside diameter cylinder with the same reaction volume because

the diameter of the cylinder is much less than the sphere. For the geometries used in this

analysis, the surface area to volume ratio of the cylinder is almost twice as great as the

sphere. Even though the average Nusselt number is twice as great for the sphere, the

convection coefficient convection is smaller because of the smaller surface area to

volume ratio of the sphere. The time constant of the spherical system versus diameter for

a 1.2 m/s air velocity, glass wall thickness of 244 µm, surface temperature of 72 °C, and

air stream temperature of 94 °C is plotted in Figure 3.21.

72

0

10

20

30

40

50

60

70

80

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0Sphere diameter (mm)

Tim

e co

nsta

nt (s

)

Figure 3.21 Plot of time constant versus diameter for spherical configuration with

constant air velocity, and wall thickness

The materials and dimensions of materials needed to optimize the reaction vessel

were found. The lumped capacitance model was used to determine all of the time

constants above. It is clear that the lumped capacitance model is an accurate model of

what is occurring as was seen in its comparison with actual experimental data. The

lumped capacitance model is an idealization that ignores conduction within the reaction

volume. This is the best case scenario because if the reaction volume was large enough

such that the lumped capacitance model was not valid, then the reaction dynamics would

be slowed by conduction within the volume. The lumped capacitance model is valid

under the following condition (Incropera and DeWitt 1996)

1.0<=k

hLBi c (3.9)

73

where Lc is the characteristic length and is given by the volume to surface area ratio of

the reaction volume. The variable k is taken to be the lumped thermal conductivity of the

reaction volume as in Equation 3.7.

The Biot number may be used to determine what reaction tube materials would

allow for quick heating. Assuming a convection coefficient of 126 W/m2K, which

corresponds to the convection coefficient for the double fan convective thermocycler

during heating for a cylindrical reaction vessel with 244 µm thick walls and an outside

diameter of 890 µm, gives a value of 0.20 W/m K as the minimum thermal conductivity

to satisfy the Biot number. Common materials used to build reaction vessels have

thermal conductivities at or above this value. The thermal conductivities of several

materials that could be used to make a reaction vessel are shown in Table 3.5 (Incropera

and DeWitt 1996; Efunda 2005).

Table 3.5 Thermal conductivities of candidate materials for PCR reaction vessels

Material Thermal Conductivity (W/mK) Polypropylene 0.17 Polycarbonate 0.20 Silicon 148 Glass (soda lime) 1.4

The thermal conductivities of the materials listed in Table 3.5 are all greater or very close

to 0.2 W/m K so any one of them could be used in the cylindrical configuration

mentioned above without causing large spatial temperature gradients within the reaction

vessel.

74

The diameter of the reaction vessel may also be varied. It is of interest to know

the largest diameter possible of the cylindrical reaction vessel such that spatial variation

in temperature does not play into the analysis. This calculation was run assuming that the

velocity of the air was 1.2 m/s, the cylinder was 39.8 mm long, the cylinder was being

heated from 72 to 94 °C, and the material was polypropylene with a wall thickness of 244

µm. The results of this analysis are shown in Table 3.6.

Table 3.6 Biot number for varying polypropylene cylindrical diameters

Cylinder Diameter (mm)

Average convection coefficient (W/m2K) k (W/mK) Bi

0.89 126. 0.26 0.11

1.0 118 0.28 0.10

2.0 82 0.42 0.10

3.0 66 0.47 0.11

4.0 56 0.50 0.12

5.0 50 0.52 0.12

6.0 45 0.54 0.13

7.0 42 0.55 0.13

The results indicate that the Biot number for the cylinders is right around 0.1. The Biot

number varies little for the diameters input above because as the convection coefficient

decreases, the value of k increases because of the greater percentage of water, and the

characteristic length increases, the net result is little change in the Biot number. These

results indicate that varying the diameter of the cylinder up to 7 mm under these

conditions has little effect on the Biot number, indicating that spatial temperature

75

gradients should be small. The problem with increasing the diameter of the tube is the

decreased convection coefficient.

The Biot number becomes larger than 0.1 if the wall thickness of the capillary is

increased. A polypropylene capillary with an air velocity of 1.2 m/s flowing past it

during heating from 72 to 94 °C, and a length of 40.0 mm was analyzed by varying the

tube wall thickness. The results of the analysis are shown in Figure 3.22.

Figure 3.22 Biot number versus wall thickness for a poly propylene capillary of outside

diameter 890 µm

Increasing the wall thickness of a polypropylene capillary under these conditions can

cause an increase in the Biot number, which is an indicator of the temperature gradient

that will be present within the reaction vessel. The reason for this is that the low thermal

conductivity of polypropylene begins to dominate the reaction dynamics because of its

thickness in relation to the water. If the capillary material is glass, which has a greater

76

thermal conductivity than water, than the Biot number decreases with increasing wall

thickness as shown in Figure 3.23.

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 390 400

Wall thickness (µm)

Bio

t num

ber

Figure 3.23 Biot number versus wall thickness for a glass capillary of outside diameter

890 µm

The decreasing Biot number means that the temperature of the reaction volume will be

nearly uniform during the thermocycling process. Using materials with a higher thermal

conductivity than water can be beneficial in making the reaction vessel more isothermal.

77

4. ANALYSIS OF MICROHEATERS FOR PCR

Another option for heating reaction volumes for PCR is the use of microheaters

(Selvaganapathy, Carlen et al. 2003; Lee, Ho Park et al. 2004). Microheaters are heaters

that have features that can be measured on the micron scale. Microsystems take

advantage of reaction volumes on the µL or in some cases nL scale to enhance the speed

of PCR (Matsubara, Kerman et al. 2004; Gulliksen, Solli et al. 2005). Reaction volumes

that have low thermal capacitance allow rapid PCR by minimizing energy storage within

the vessel. The microheater is typically tightly coupled to the reaction volume (Yoon,

Lee et al. 2002), allowing heat transfer to occur rapidly, unlike Peltier devices which are

typically covered with a ceramic layer, and in some cases have metal blocks mounted on

top of them.

The work of other groups in this area was studied and a survey of that work is

described in Section 4.1. Microheaters were designed and fabricated using standard

MEMS techniques, as discussed in Section 4.2 . A control system, discussed in Section

4.3, for temperature cycling with the microheater was built and tested.

4.1 Synopsis of work done on microheaters by other groups

SU-8 is one option for creating microchambers on top of microheaters. It has

been used as a mold for PDMS microchambers (Zhao, Cui et al. 2002). El-Ali and others

have recently incorporated a thin film platinum heater with an SU-8 reaction volume to

achieve heating and cooling rates five times faster than the fastest macroscale device and

power consumption 126 times less (El-Ali, Perch-Nielsen et al. 2004). The device

created by this group consists of an SU-8 based PCR chamber that is affixed to a 1 mm

78

thick glass substrate with a thin film platinum heater deposited on it. There are 83

individual heater elements, each element is 20 µm wide and 8.3 mm long with spacing

between heater elements of 100 µm. The heaters extend beyond the boundaries of the

reaction chamber to minimize cold wall effects. The glass also has an independent

thermometer separate from the heater itself. The resistance of the platinum thermometer

changes linearly with temperature making it possible to determine the temperature by

measuring the resistance. The reaction chamber is 7 x 7 x 0.4 mm, giving a total reaction

volume of 20 µL. Atop the reaction chamber a 0.5 mm glass piece is bonded with SU-8

utilizing SU-8 to SU-8 bonding. The lid has two small openings in it, through which the

reaction chamber may be filled. It was not mentioned how these holes were sealed

during cycling. The chip is cooled via natural convection and through a heat sink

connected to the lower surface. A Labview PID controller was used to control the

heating and cooling of the chip. A schematic of the device is shown in Figure 4.1 (El-

Ali, Perch-Nielsen et al. 2004).

Figure 4.1 Heater with integrated well as designed by El-Ali et al.

79

El-Ali et al. used standard MEMS techniques to fabricate the device shown above.

The heater electrodes and the thermometer were patterned with lift-off on the glass

substrate. The platinum was deposited via e-beam evaporation using a 100 Å titanium

layer for adhesion. A 5 µm SU-8 layer was spun onto the wafer. The pattern was soft

baked and patterned with photolithography. The purpose of this layer was to protect the

electrodes from the PCR solution. The 400 µm deep reaction vessel was defined with

two spins of 200 µm each. A soft bake step was used in between each spin. The glass lid

was next bonded to the SU-8 reaction vessel using SU-8 to SU-8 bonding by first

applying a thin coat of SU-8 to the glass lid, soft baking the lid with SU-8 on it,

positioning the lid on top of the SU-8 section, soft baking, exposing with UV light, and

finally another bake. It was necessary for the researchers to treat the SU-8 surfaces of the

reaction chamber with the silanizing agent dichlorodimethylsilane to make it PCR

compatible. The dichlorodimethylsilane makes the surface PCR compatible by inhibiting

adsorption of DNA and polymerase to the surface of the chamber (Kopp, de Mello et al.

1998). The final step in the production of the chip was drilling two holes in the glass lid

for filling of the reaction vessel (El-Ali, Perch-Nielsen et al. 2004).

Silicon has also been tested as a material for microchambers (Daniel, Iqbal et al.

1998; Nagai, Murakami et al. 2001; Matsubara, Kerman et al. 2004). Zhao et al.

designed a PCR microchip with an integrated heater utilizing silicon as the material for

the reaction vessel (Zhao, Cui et al. 2003). The device developed by this group was an 8

x 4 mm chip. The reaction vessel is 4 x 2 x 0.27 mm, giving a total reaction volume of 2

µL. The reaction vessel was manufactured on one side of the chip while the thin film

platinum heater was deposited on the opposite side of the chip. The heater was separated

80

Figure 4.2 PCR reaction vessel designed by Zhao et al. (Zhao, Cui et al. 2003).

from the reaction volume by a distance of only 2 µm. Two

channels were etched on opposite sides of the reaction

vessel to provide a method for filling the vessel with PCR

reagents. The chip was anodically bonded with a 0.2 mm

thick piece of borosilicate glass. Two holes were drilled in

the glass and were aligned with the chip to allow filling.

A schematic of the setup is shown in Figure 4.2. The

device consists of a heating power supply circuit,

temperature controller, keypad, and display. The heater was controlled with a

microcontroller and a PID scheme. The heater power was controlled with pulse width

modulation at a frequency of 19.5 kHz (Zhao, Cui et al. 2003).

The device was fabricated using a 270 µm double-side polished silicon wafer.

Prior to etching, silicon nitride was deposited on each side of the wafer to a thickness of 2

µm. The reaction vessel was wet etched with KOH utilizing the silicon nitride as a stop

layer. A 100 nm oxide layer was deposited after the wet etch to provide a suitable

surface for the PCR reaction. Two 1 mm length channels were etched on each side of the

reaction vessel to allow filling. The platinum heater and temperature sensor were

patterned on the backside of the wafer using standard photolithography and lift off. The

chip was next mounted on a PCB board with patterned electrodes to connect to the heater

(Zhao, Cui et al. 2003).

Researchers have turned to plastic microchambers for use with microheaters

(Giordano, Ferrance et al. 2001; Liu, Rauch et al. 2002). Zou et al. have taken this

approach to designing a PCR microchip with integrated heater. The device consists of

81

multiple silicon blocks with a thickness of 680 µm bonded to a 500 µm PCB substrate

with a heat sink attached to the bottom. Heaters and temperature sensors are fabricated

on the underside of the silicon blocks. The reaction chambers were thermally molded in

plastic with a silicon plate etched with KOH. The type of plastic used was not specified.

The 4 x 4 array of reaction chambers are affixed to the top of the heaters with thermal

grease and mechanical means. The reaction chambers are 400 µm deep, are

approximately 5 mm square, and have a wall thickness of 100 µm yielding a reaction

volume of 20 µL. A schematic of the device is shown in Figure 4.3 (Zou, Miao et al.

2002).

Figure 4.3 Microheaters with integrated reaction vessels as designed by Zou et al.

The heater array manufactured by Zou used standard MEMS techniques. The

process began with a thermal oxide deposition on the top of the wafer. The heaters and

temperature sensors were manufactured by depositing the aluminum to a thickness of 2.4

µm and then patterning. Plasma enhanced chemical vapor deposition was next used to

deposit a silicon nitride layer on top of the heater elements. This layer was patterned to

allow connections with some of the underlying metal components of the heater with the

PCB board. The next step for the designers was to deposit and pattern the under bump

metal (UBM) layers. This consisted of a Ti/Ni/Au multilayer. Soldering balls were next

printed onto the UBM surfaces. The PCB was produced with electrical connections for

Silicon block with heater on bottom

Heat sink PCB

82

the heaters and temperature sensor. The silicon blocks were placed on top of the PCB

and soldered using temperature and pressure in a flip chip bonding technique. The PCB

was next mounted to a heat sink with electrically non-conductive epoxy. The researchers

used polypropylene tape or bonded a thin sheet of plastic to the top of the reaction vessels

with epoxy or with solvent bonding. A schematic of the process sequence is shown in

Figure 4.4 (Zou, Miao et al. 2002).

Figure 4.4 Flip chip design process used by Zou et al. for fabrication of PCR

microheaters

4.1.1 Comparison of designs

The aim of a microscale device is to complete the PCR reaction as quickly as

possible with little power consumption. Table 4.1 at the end of this section summarizes

the operating parameters of the heaters designed by the researchers discussed above. The

maximum heating rate of the device manufactured by El-Ali was approximately 50 °C/s,

with a cooling rate of about 12 °C/s at a location on the heater surface. The power

consumption of the heating device was 6.3 W at 94 °C (El-Ali, Perch-Nielsen et al.

2004). The device manufactured by Zhao achieved heating rates of 15 °C/s and cooling

83

rates of 10 °C/s at a location on the heater. This device was shown to be able to

successfully complete a PCR reaction in 15 minutes (Zhao, Cui et al. 2003). Zou’s group

fabricated a device capable of a 42 °C/s heating rate and a 27 °C/s cooling rate with a 15

µL sample loaded in the reaction volume measured at the surface of the heater. Zou’s

device used, on average, only about 0.9 W per chamber . Zou’s device was capable of

successfully completing a PCR reaction with a 320 bp piece of plasmid DNA in a time of

15 minutes (Zou, Miao et al. 2002).

The manufacturing approach taken by El-Ali et al utilized SU-8 as the reaction

volume. SU-8 is a polymer. El-Ali found that the reaction volume needed to be silanized

in order to be PCR compatible. It was also necessary for Zhao’s group to add an oxide

layer to the silicon to make it PCR friendly. The reaction chamber used by Zou’s group

was made of an unspecified form of plastic and was formed with thermal molding. Zou

was the only researcher who did not report the need to passivate the surface of the

reaction volume.

Zou chose aluminum as the metal for his heater. The choice of aluminum is an

interesting one. The other two researchers presented chose platinum as the metal for the

heaters. Zou reported that he was using aluminum thicknesses upward of 2.5 µm. Zou

did not report how this was done

Zou’s device was the most ready for mass production. Zou incorporated an array

of heaters in his design and had a thermal molding process for producing reaction

chambers. Thermal molding is faster than building reaction vessels out of SU-8 or

etching them out of silicon, processes used by El-Ali and Zhao. Zou also utilized flip

chip bonding, which enabled him to quickly attach his heater assembly to the PCB board.

84

Zou’s device was the only one to utilize separate reaction chambers from the

heaters. El-Ali had his reaction volume attached to the heater and Zhao had his heater

manufactured on the same wafer as the reaction volume. This strategy by El-Ali and

Zhao was a poor one. Their heating and cooling rates were no greater than Zou’s. Zou’s

heater may be reused, however, by simply throwing away the reaction volumes and

reusing the heaters. Metal deposition is an expensive and time consuming process.

Platinum prices are approximately $40/gram so it is not a good idea to throw heaters

made of platinum away. Zou had a good strategy when considering the reuse of his

mechanism.

A summary of the designs is shown in Table 4.1. The two conventional choices

for PCR, the Peltier device, Techne Techgene studied earlier, and the LightCycler, a

convection device, use large amounts of power and have heating and cooling rates slower

than microheater systems. Neither of the traditional devices require surface treatments.

Two of the microheater devices required surface treatments due to the higher surface area

to volume ratio of these devices.

Table 4.1 Comparison of thermocycling devices (Zou, Miao et al. 2002; Zhao, Cui et al.

2003; El-Ali, Perch-Nielsen et al. 2004; LightCycler 2005)

Power Consumption

(W)

Heating Rate (°C/s)

Cooling Rate (°C/s)

Number of Reaction

Chambers

Reaction Volume

(µL)

Surface Treatment Necessary for

Biocompatibility?

Time to Complete

PCR Reaction (minutes)

Traditional Peltier Device 264 (max)

3.6 (max heating block)

2 (max heating block) 20 20 No 150

Traditional Convective Device 800 (max)

5.5 (average in capillary)

5 (average in capillary) 20 10 No 30

El-Ali's Device 6.4 (@ 94 °C)

50 (max on heater)

30 (max on heater) 1 20 Yes N/A

Zhao's Device N/A

15 (max on heater)

10 (max on heater) 1 2 Yes 15

Zou's Device 0.9 (Average)42 (max on

heater)27 (max on

heater) 1 20 No 15

85

4.2 Microheater design

There are several key aspects in designing a microheater for PCR. The first items

to be kept in mind are what will the power source be, how much current will the control

system be able to handle, how big should the heaters be, what material should the heaters

be made of, on what substrate will the heaters be fabricated on, and what heat flux is

desired.

The power required by the heater in our case was an important question. The

heater that we decided to design was to have platinum heating elements along with a

platinum temperature sensor. The heater needed to be powerful enough to increase the

sample temperature by greater than 6 K/s in order for it to heat faster than commercially

available devices such as the LightCycler which is capable of heating the reaction volume

at 6 K/s (LightCycler 2005). A 2-D finite element heat transfer package was used to

understand the power requirements needed by a microheater. A 20 W/cm2 heater was

desired, so this power requirement was examined using the 2-D package. The material

that looked promising for a reaction vessel was polycarbonate. Polycarbonate can be

easily machined, molded, and bonded (Martin, Matson et al. 1998; Klintberg, Svedberg et

al. 2003). The first case modeled was a polycarbonate well machined into a 2 mm thick

substrate. The well was 3.2 mm in diameter and 1.5 mm deep. The model was arranged

such that 10 µL of water was in the well. This meant that the height of the water in the

well was 1.2 mm. The model was axisymmetric and was done with a cylindrical

coordinate scheme. The heat flux across the centerline was considered to be zero. A

schematic of the model is shown in Figure 4.5.

86

Figure 4.5 Schematic of 2-D model used in analysis of heating power

The model was run to determine the heating characteristics of the fluid in the chamber

assuming that the water in the chamber is stationary. The results of that analysis are


0

50

100

150

200

250

300

350

400

0 1 2 3 4 5

Time (seconds)

Tem

pera

ture

(°C

)

Bottom of water at midlineCenter of water at midlineTop of water at midline

Figure 4.6 Response of 3.2 mm diameter polycarbonate chamber of thickness 2 mm

with 10 µL of water in it and a heat flux of 20 W/cm2

87

The three curves in Figure 4.6 are for three different positions in the reaction

vessel. Figure 4.5 shows the reaction vessel with the midline at the center. The three

lines in Figure 4.6 represent three positions along that axis in the water. The portion of

the chamber in contact with the water heats up rapidly while the heating rate decreases as

the distance from the heat flux increases. The heat flux of 20 W/cm2 heats the bottom

portion of the chamber to 88 °C in one second. This is satisfactory, but it is desired to

heat the water in the reaction vessel rapidly also. This model has a polycarbonate

thickness of 500 µm between the heat flux and the water with a water thickness of 1500

µm on top of it.

A polycarbonate chamber fashioned out of 2.0 mm section with a 1.5 mm deep

hole and a diameter of 5.0 mm was analyzed with the 2-D finite element package to

compare the performance with the previous example. This geometry allowed 10 µL of

water to occupy the reaction chamber at a height of 500 µm as opposed to 1500 µm in the

previous case. A schematic of the model is very similar to Figure 4.5 except that the

dimensions are slightly different. The results of the analysis are shown in Figure 4.7.

88

0

50

100

150

200

250

300

350

400

0 1 2 3 4 5Time (seconds)

Tem

pera

ture

(°C

)Bottom of water at midlineCenter of water at midlineTop of water at midline

Figure 4.7 Response of 5 mm diameter polycarbonate chamber of depth 2 mm with a 10

µL sample of water in it and a heat flux of 20 W/cm2

The average heating rate for the water at the top of the water sample for the first second is

10 °C/s and increases to 28.0 °C/s over a two second duration. These heating rates for

the sample were considered acceptable and are comparable to those achieved by the

researchers in section 4.1. It is clear that the bottom of the reaction chamber reached over

150 °C within two seconds of heating with this heat flux. This excessive heat is a

problem that will be handled by the control system. A short burst of high heat flux could

be supplied by the heater and after this the heater power could be reduced. As a result of

this modeling, a heater flux of 20 W/cm2 was chosen.

The heater flux may be related to the size of the heater and the resistance of the

heater. The power output by electricity flowing through a resistor is given by

89

IVP = (4.1)

The current is given by Ohm’s Law

RVI = (4.2)

Substituting Equation 4.2 into Equation 4.1 gives an expression relating the heater

resistance, the applied voltage, and the power.

RVP

2

= (4.3)

Now the desired resistance of the heater may be found, knowing the desired power and

the voltage. The desired input voltage was 10 V based on safety and practical

considerations of available power supplies. Two square heaters were desired, a 5 mm

and a 2.5 mm heater. The power dissipated by these two heaters considering that a 20

W/cm2 heating flux was desired was 5 and 1.25 W respectively. This gives values of 80

Ω for the 2.5 mm square heater and a resistance of 20 Ω for the 5 mm square heater.

The next step in the design of a platinum resistance heater is to determine the

heating element orientation such that the heating elements fit well within the allotted

space and the resistance requirement is satisfied. In order to design the heater elements,

the resistivity of deposited platinum must be known. The resistance of a conductor is

given by the following formula

wtLR ρ

= (4.4)

where ρ is the resistivity of the material (typical units are Ω-m), L is the length of the

wire, and w and t are the width and thickness of the wire respectively, corresponding to

the wire cross section. It was a given that e-beam evaporation followed by lift-off would

be used for the process sequence since there was uncertainty about obtaining a platinum

90

target for sputtering. Platinum heaters were initially manufactured using a resistivity

value for platinum of 1 x 10-7 Ω-m (Winter 2005). This resistivity is a bulk value for

platinum. It was discovered that platinum deposited on top of a thin chrome adhesion

layer had a higher resistivity than that listed above after experimentation. The actual

resistivity of a platinum wire of thickness 0.1 µm, a width of 50 µm and a length of 4.7

mm on a chrome adhesion layer of 150 Ǻ was 4.4 x 10-7 Ω-m as measured with a

multimeter without any heat treatment of the film. This value of resistivity was therefore

used in subsequent design calculations. Researchers have reported that this effect is due

to a large number of open and closed voids in the platinum film causing the density of the

platinum to decrease with a consequent increase in resistivity (Lourenco, Serra et al.

1998; Giani, Mailly et al. 2002).

The information for the resistivity and power requirements of the heater was used

to design two square heaters of size 2.5 and 5 mm. It was decided that the platinum

thickness would be 0.1 µm since thicker platinum layers have higher stress in them that

may cause them to crack (Rottmayer and Hoffman 1971). The 5 mm heater was designed

with 39 individual 5 mm long heater elements in parallel. The elements had a cross

section of 0.1 x 30 µm. This gave an element resistance of 725 Ω and a heater resistance

of 18.6 Ω (725 Ω/39). Masks were designed for fabrication of the heaters. Schematics of

the 5 mm heater are shown in Figures 4.8 and 4.9.

91

Figure 4.8 Schematic of 5 mm heater. All dimensions in µm.

The 39 parallel heater elements are visible in Figure 4.8. A platinum temperature sensor

is located in the center of the device. The intent of the design was to deposit a 150 Ǻ

chrome adhesion layer followed by 1000 Ǻ of platinum. The large connectors at each

end of the heater have a chrome adhesion layer deposited on top of them followed by a

layer of gold. Gold was chosen as the material for the leads because of its low resistivity

of 0.22 x 10-7 Ω-m. Low resistivity and large cross sections were designed to keep power

dissipation in this area to a minimum. The leads for the 5 mm heater were 900 µm wide

as shown in Figure 4.8. The resistance of the lead at the top of the heater was 4.5 Ω

considering that it is 9.4 mm long, 900 µm wide, and has a thickness of 0.1 µm. The

temperature sensor is clearly visible in Figure 4.9. The temperature sensor was designed

with a resistance of 617 Ω.

92

The 2.5 mm heater was designed with a heater resistance of 80.6 Ω. There were 9

heater elements each with a resistance of 725.4 Ω considering a resistivity for platinum of

4.4 X10-7 Ω-m, a line width of 30 µm, and a deposition thickness of 0.1 µm. Schematics

of the 2.5 mm heater are shown in Figures 4.10 and 4.11.

Figure 4.9 Exploded view of 5 mm heater with some elements deleted for clarity.

Center portion is the temperature sensor. All dimensions in µm.

93

Figure 4.10 Schematic of 2.5 mm heater. All dimensions in µm.

The 2.5 mm heater was similar to the 5 mm heater in that the design intent was to deposit

a 150 Ǻ layer of chrome followed by 0.1 µm of platinum. The gold leads would then be

formed by putting a chrome gold layer on top of the platinum in the lead areas. In this

design, the heater elements are not single lines, in this case. The elements double back

on themselves, thereby increasing the resistance. An exploded view of the heater

elements and temperature sensor are shown in Figure 4.11.

94

Figure 4.11 Exploded view of 2.5 mm heater. Center portion is the temperature sensor.

All dimensions in µm.

The temperature sensor in this case was designed with a resistance of 435 Ω. The heater

elements were staggered with the part of the heater that wraps back on itself on

alternating sides of the heater. The purpose of this was to give more uniform heating.

Masks were also created for the second step of photolithography which was the

deposition of the gold leads. These masks are shown in Figures 4.12 and Figures 4.13.

All of the masks shown are actually inverted from the actual masks. The actual masks

had the black area transparent and the surrounding area opaque.

95

Figure 4.12 Gold lead mask for 5 mm heater

Figure 4.13 Gold lead mask for 2.5 mm heater

4.2.1 Electronic control system design

The temperature of the heaters needed to be controlled in order to run the PCR

reaction. The options that were considered were PWM and a continuous controller that

utilized a Wheatstone bridge. The continuous controller configuration was chosen

because it made use of the heater element as the temperature sensor for the control

system. A Wheatstone bridge is a configuration of resistors that is commonly used to

measure strain gages. This configuration has been used by other researchers for

96

temperature control and is documented in the literature (Benammar and Maskell 1989).

A diagram of the circuit used to control the microheaters is shown in Figure 4.14.

Figure 4.14 Circuit schematic for microheater control

The main function of the control system is to feed current to the heater (Rh) causing its

temperature to be controlled at a fixed value for a given set of values of the variable

resistors and the digital potentiometer. The circuit works by amplifying the imbalance in

the bridge and supplying more current when the error is larger. The fact that the heater

element resistance follows a known function of temperature allows the heater element to

act as the sensing element as well as the final control element. The DS1267 is a digital

potentiometer that is controlled by a computer via a three wire serial interface. The

purpose of the digital potentiometer is to provide a setpoint to the control circuit. The

7805 is a power regulator. It takes the input voltage and regulates it to 5 V in order to

power the digital pot. The purpose of the L1494 operational amplifier is to amplify the

97

differential input in order to open or close the transistor. The purpose of the 10 kΩ

resistor around the transistor is to allow a small amount of current to flow through the

bridge at startup to initiate the control process.

The control system can be calibrated to control the temperature of the heater. The

resistance of the platinum heater increases with temperature. This fact can be used to

control the circuit. When the temperature of the heater has reached the proper

temperature, the voltage difference across the bridge should be zero. The voltage

difference across the bridge can be read at the input to the operational amplifier. This

information can be used to set the 0-10 kΩ resistor and the 0-1 kΩ resistor for a particular

heater. At bridge equilibrium the following expressions hold (Alciatore and Histand

2003)

02

2

41

1 =+

−+ hBB RR

RRR

R (4.5)

02

2

41

1 =+

−+ hEE RR

RRR

R (4.6)

where R1 and R2 are the resistances as shown in Figure 4.14. RhB refers to the resistance

of the platinum heater at the top of the PCR operating range, in this case 100 °C. RhE

refers to the resistance of the platinum heater at ambient temperature, 21 °C. R4B refers

to the resistance of the leg of the bridge containing the digital potentiometer when the

heater was at its upper setpoint of 100 °C, while R4E refers to the resistance of that leg at

ambient temperature. The potentiometer was set at 20 kΩ for operation of the heater at

its upper setpoint and 0 kΩ for operation at ambient temperature. R4 is given by the

following equations

98

p

SB

R

RR1

200001

14

++= (4.7)

p

SE

R

RR1

10001

14

++= (4.8)

where Rs is the resistance of the manual potentiometer in series with the parallel

combination of the digital potentiometer and Rp, the resistance in parallel with the digital

potentiometer. These equations may be solved simultaneously to obtain the values of the

manual potentiometers.

The onboard temperature sensor was used as a redundant method to determine the

temperature of the heater. The temperature sensor contact pads were hooked up to a

simple circuit that used a 500 Ω resistor and a voltage measurement, as shown in Figure

4.15, to determine the resistance of the sensor (Rt), and then the temperature could be

inferred from the calibration of resistance versus temperature.

Figure 4.15 Schematic of circuit used to measure temperature of platinum temperature

sensor

VRT

99

The voltage was measured across the platinum thermometer, Rt, as shown in Figure 4.15.

The current through the circuit could be found with the following equation

500

5R

VVI RT−= (4.9)

where VRT is the voltage across the platinum thermometer. The resistance of the

platinum thermometer could be found from Ohm’s Law knowing the current.

IVR RT

t = (4.10)

4.3 Microheater fabrication

The microheaters were fabricated using standard photolithographic methods.

Photolithography was performed using positive photoresist. After exposure, a

chlorobenzene soak and drive in period was used to form an outer layer on the surface of

the photoresist that would develop slower(Halverson, MacIntyre et al. 1982). E-beam

evaporation was used to deposit the chrome and then platinum (chrome adhesion layer)

after photolithography. Lift-off was used to form the heater pattern.

4.3.1 Microheater fabrication equipment and techniques

The masks for the heaters were designed in AutoCad and then converted to Adobe

Illustrator. The masks were designed as negatives, meaning that areas that were black in

the original would be clear in the final product and vice versa. The masks were printed

by RGM Graphics, Inc., of Bethesda Maryland. They were printed on a 100 µm thick

mylar sheet. The masks were setup to be printed on a standard sheet with dimensions

215.9 X 279.4 mm. A 3000 dpi printer was used to print the masks.

100

The wafers used in the fabrication process were 700 µm thick, 3 inch diameter

soda lime glass wafers obtained from Valley Design Corp. An optically clear substrate

was desired because some detection schemes involve shining light on the reaction volume

from one side and detecting fluorescent response on the other side. The wafers were first

rinsed with acetone, methanol, isopropanol, and finally with deionized water. Hot plates

were used to dehydrate the wafers and to harden the photoresist. The hot plates used

were Thermolyne type 1900 with Taylor 9940 temperature controllers.

The photoresist application, exposure and development required several items.

S1813 photoresist from Rohm and Haas was spun onto the wafers using a Headway

Research Inc. spinner at 4000 RPM for 40 seconds. Exposure was performed using the

masks from RGM Graphics and the Karl Suss MJB3 type 100DU030 photolithography

machine. The wafers were exposed with a UV light intensity of 8 mW/cm2 at 365 nm for

25 seconds. Chlorobenzene was obtained from JT Baker, product number 9179 and was

used to profile the photoresist. An oven was used to drive in the chlorobenzene. The

oven used was from Blue M, model OV-8A. Development of the photoresist took place

in a 3:1 solution of AZ400K:water. The AZ400K photoresist developer was obtained

from Microposit.

Several components were necessary for the metal deposition. A buffered oxide

etch (BOE) was necessary to prepare the surface of the glass. The buffered oxide etch

(BOE) used was BOE Improved supplied by Transene Company, Inc. The BOE was

carried out in a plastic container using plastic wafer holders. The chrome used was stock

chrome while the gold and platinum were obtained from Kurt J. Lesker Company. The

gold and platinum were 99.99% pure and came in 6.35 mm diameter 6.35 mm long

101

cylinders. The thermal evaporator used to deposit the gold was from Metra Inc., model

TEBC-22-26. The e-beam evaporation unit was a Temescal, model 1800. Profiles of

metallized features were taken with a Tencor Alphastep profilometer.

4.3.2 Microheater fabrication sequence

A simplified process sequence is shown in Figure 4.16.

Figure 4.16 Simplified fabrication sequence for microheaters

The wafers were cleaned with acetone, methanol, isopropanol, and DI water. The

wafers were next placed on a hot plate at 120 °C for two minutes to dehydrate the wafer

to prepare it for photoresist application, shown in step 1 in Figure 4.16. A 2 minute soft

bake was done at 120 °C after the application of the photoresist to prepare it for exposure.

The wafer was then exposed and soaked in chlorobenzene for 15 minutes, corresponding

to steps 2 and 3 above. The purpose of the chlorobenzene is to form an undercut

102

structure in the photoresist. The chlorobenzene penetrates into the photoresist to a depth

determined by the length of exposure to it. It removes residual solvent and low molecular

weight resins. This causes the affected region to develop more slowly then the region

underneath it (Halverson, MacIntyre et al. 1982). This aids in lift off by forming a

discontinuous metal structure that can be easily lifted off. A blown up picture of

chlorobenzene treated photoresist profile after development corresponding to step 4 is


Figure 4.17 Effect of chlorobenzene on photoresist development

The chlorobenzene soaked wafer was taken out of the solution, blow dried with N2, and

inserted into the oven at 90 °C for 15 minutes to drive the chlorobenzene into the

photoresist. The wafer was then developed in the AZ400K solution and optically

inspected under the microscope to ensure that complete development had taken place.

The wafer was metallized with chrome and platinum. The wafer was hard baked

at 120 °C for 5 minutes to strengthen the resist matrix prior to metallization. A 1 minute

BOE was performed to roughen up the exposed glass surface, allowing the chrome to

adhere better to the surface, corresponding to step 5 in Figure 4.16. The wafer was

placed into the e-beam evaporation chamber and 150 Å of chrome was deposited

followed by 1000 Å of platinum. The platinum and chrome deposition rates were kept

below 5 Å/s. The wafer was removed from the e-beam evaporator and placed in an

103

acetone bath to lift off the unwanted metal and photoresist. The wafer was soaked for

approximately 1 hour. The wafer was then removed, rinsed with DI water, blow dried

with N2, and optically inspected corresponding to step 7 in Figure 4.16.

The gold leads were then deposited on the heaters. The wafer was cleaned, blow

dried, dehydrated, and had photoresist applied to the metallized surface. A 2 minute soft

bake was done at 120 °C as in the previous step. The trace mask was then aligned to the

metallized area using the features as guides. Exposure was carried out on the wafer and

the chlorobenzene soak and drive in step were carried out. Development proceeded as

described above. The wafer was then hard baked at 120 °C for 5 minutes. The wafer was

placed in the thermal evaporator and 150 Å of chrome was deposited followed by 1000 Å

of gold. The wafer was removed from the thermal evaporator and soaked in an acetone

bath for 1 hour followed by rinsing with DI water. The wafer was then blow dried with

N2 and optically inspected under the microscope.

4.3.3 Microheater fabrication results

The microheaters suffered from some delamination problems after liftoff in the

acetone solution. Two 2.5 mm square heaters and two 5 mm square heaters were

manufactured on a wafer. The 2.5 mm heaters did not turn out at all due to a mask

problem that made it impossible to align both the 2.5 and the 5 mm heaters. 5 mm

heaters were the only ones produced for this reason. Typical delamination of heater

elements is shown in Figure 4.18.

104

Figure 4.18 Delamination of heater elements from soda lime glass substrate

The delamination shown in Figure 4.18 is for a heater with a 150 Å titanium adhesion

layer with 1000 Å of platinum deposited on top of it. This type of delamination was

common for the case where a chrome adhesion layer was used also. It was typical that

only one heater would turn out with no discontinuities in the heater elements due to

delamination out of four heaters manufactured. A discontinuity in a heater element refers

to a heater element with a section missing due to delamination or photolithography

problems. Even heaters that did not suffer from discontinuities in any heater elements

had imperfections as shown in Figure 4.19.

105

Figure 4.19 Heater elements from a heater with no discontinuities

The image shown in Figure 4.19 is from a heater with no discontinuities in any of the

heating elements. It is clear that the heater elements have pits, particularly at the edges of

the elements. The feature size of the heater elements was 30 µm. Elements with feature

sizes greater than 30 µm had fewer defects. The comparison of 50 µm wide temperature

sensor elements with 30 µm wide heater elements is shown in Figure 4.20.

Figure 4.20 Comparison of 30 µm heater elements with 50 µm temperature sensor

Heater elements

Temperature sensor

106

The 50 µm wide heater element clearly has fewer defects than the heater elements. The

result of this problem could be found in measuring the resistance of the temperature

sensor and the heater and comparing it with the design resistance. The design resistance

of the 5 mm platinum heater was 20 Ω while the design resistance of the temperature

sensor was 617 Ω. The actual resistance of the heater immediately after fabrication was

18.2 Ω and the temperature sensor resistance was 470 Ω. The design resistance was

calculated considering that the resistivity of platinum was 4.4 x 10-7 Ω-m with a

deposition thickness of 1000 Å. The actual heaters produced had a platinum deposition

thickness of 817 Å with a chrome adhesion layer underneath of it. Profilometry revealed

that this was indeed the case. Observation under a microscope indicated that the width of

the elements matched the design intent. The effective resistivity of the platinum

considering these factors was 3.6 x 10-7 Ω-m for the heater itself and 2.7 x 10-7 Ω-m for

the temperature sensor. The fact that temperature sensor resistivity was much lower than

the heater resistivity was attributed to the imperfections in the heater elements which

impede current flow through the device.

4.4 Microheater heat treatment and calibration

It was found that the resistance of the heaters and temperature sensors were

changing with time making it very difficult to control the heaters and making temperature

measurements with the platinum thermometer problematic. A study was done to

determine the effect of heat treatment on the heaters in an effort to stabilize the heater and

temperature sensor resistance. The resistance of four 5 mm heaters was tracked during

this study. The test consisted of baking the heaters in an oven at temperatures ranging

107

from 200 to 265 °C. The resistances of the heaters and temperature sensors were

measured using a multimeter. The resistances were measured at the points indicated in

Figure 4.21.

Figure 4.21 Points of resistance measurement for 5 mm microheater

The resistance of the thermometer was measured between the thermometer contact pads.

The heater consists of two heater sections in parallel if the heater ground pads are linked

together. R1 was measured between the bottom right contact pads while R2 was

measured between the bottom right contact pad and the left most contact pad. The

resistance of the heater is the parallel combination of R1 and R2 and is given as Rh. The

results for one of the heaters are summarized in the following table.

Table 4.2 Result of heat treatment on heater and temperature probe

Temperature (°C) Time Total elapsed time (h) R1 (Ω) R2 (Ω) Rh (Ω) Rt (Ω)21 0.0 0.0 N/A N/A 18.2 470

200 1.5 1.5 N/A N/A 16.7 387225 1.5 3.0 32.0 35.3 16.8 380225 0.8 3.8 31.9 35.4 16.8 376225 1.5 5.3 32.0 35.6 16.8 375265 1.0 6.3 32.2 35.6 16.9 371265 1.0 7.3 32.4 35.6 17.0 370265 1.0 8.3 32.6 36.0 17.1 369265 1.0 9.3 32.5 36.0 17.1 367225 1.5 10.8 32.4 36.5 17.2 366205 3.0 13.8 32.4 35.8 17.0 366

Heater ground pad

Heater ground pad

108

The results of the heat treatment on the 5 mm heater indicate that the resistance of

the temperature probe decreased by 22 % and the resistance of the heater decreased by

7 %. Over the last 5.5 hours of heat treatment the resistance of the heater changed by

only 0.1 Ω and the resistance of the thermometer change by only 1 Ω indicating that the

resistances had stabilized for heat treatment at these temperatures. The effective

resistivity of the heater after heat treatment was 3.2 x 10-7 Ω-m while the effective

resistivity for the temperature sensor was 2.1 x 10-7 Ω-m. It has been reported that after

heat treatment up to a temperature of 1000 °C the resistivity of a 110 µm thick platinum

film on a titanium adhesion layer reached 1.4 x 10-7 Ω-m (Lourenco, Serra et al. 1998).

The resistance of the heater was measured at ambient and at 100 °C in an oven in

order to calibrate the electronic control system. The heater used for this test was another

one of the heaters that had been heat treated but is not the same one listed in Table 4.2.

These temperatures define the limits of operation for the device. The temperature versus

resistance data was plotted for the heater and is shown in Figure 4.22.

y = 51.737x - 1035.5R2 = 0.9963

0

20

40

60

80

100

120

20 20.2 20.4 20.6 20.8 21 21.2 21.4 21.6 21.8 22Resistance (Ω)

Tem

pera

ture

(°C

)

TemperatureLinear (Temperature)

Figure 4.22 Temperature versus resistance with linear fit for microheater

109

The resistance of the heater increased from 20.4 Ω at ambient temperature to 21.9 Ω at

100 °C, a difference of 1.5 Ω or an increase in 7.4% from its ambient value. The

calibration plot for the platinum thermometer is shown in Figure 4.23.

y = 2.4215x - 855.48R2 = 0.9992

0

20

40

60

80

100

120

360 365 370 375 380 385 390 395 400Resistance (Ω)

Tem

pera

ture

(°C

)

Figure 4.23 Temperature versus resistance graph with linear fit for microheater

temperature sensor

The resistance of the temperature sensor increases from 362.6 Ω at ambient temperature

to 394.7 Ω at 100 °C. This is a difference of 32.1 Ω or a percent increase of 8.9 %. The

slope of the calibration curve for the heater is much steeper than the slope of the curve in

Figure 4.23. The percent change in resistance for each case is nearly identical but the

resistance range over which it happens for the heater is much smaller than that for the

temperature sensor.

110

Heater recalibration was redone in the oven after 1 week of use to determine if the

resistance of the temperature sensor was remaining constant. The result of that test is


0

20

40

60

80

100

120

360 365 370 375 380 385 390 395 400

Resistance (Ω)

Tem

pera

ture

(°C

)

First calibrationtemperatureSecond calibrationtemperature

Figure 4.24 Comparison of initial calibration of temperature sensor and after 1 week of

use

The results show that the temperature sensor was slightly out of calibration after 1 week

of use, especially at lower temperatures. At temperatures higher than 60 °C there was

agreement between the first and second calibrations.

4.5 Microheater testing

The microheaters were tested for heating rate, cooling rate, and power

consumption under two conditions. The first condition was with fan cooling. The fan

was turned on during cooling from 94 to 54 °C, but was otherwise off. The second

111

condition was with the cooling fan and a heat sink placed underneath the heater to speed

cooling.

The heater circuit was powered with a Mastech HY1803D power supply. The

cooling fan mounted on top of the test fixture was a RadioShack brushless, 12 VDC

blower fan, part number 273-199 as shown in Figure 4.25.

Figure 4.25 Microheater test fixture

The circuit that was used to measure the temperature of the heater, and the blower fan

were powered by an ATC power supply, model ATC-230U. The fan was supplied with

12 V, while the temperature sensing circuit was supplied with 5 V. The variable resistor

was controlled with the Instrunet 100 data acquisition system. The temperature of the

heater, heater voltage, period, and three other thermocouple temperatures were recorded

once every second by the Instrunet 100.

The second configuration of the microheater tested was with a heat sink, obtained

from RadioShack, part number 276-1368, underneath the microheater. The purpose of

the heat sink was to determine if it would help in cooling the device from 94 to 54 °C.

The microheater test fixture with heat sink is shown in Figure 4.26.

Blower fan

112

Figure 4.26 Microheater test fixture with microheater mounted on heat sink

The power calculations were made by measuring the voltage across the heater

using an inverting operational amplifier that would multiply the input voltage by -0.22

and then output that voltage to the data acquisition unit. The resistance of the heater

could be determined by knowing the calibration curve for the heater and the temperature

of the heater. The temperature of the heater was determined by the platinum

thermometer. The power supplied to the fan was significant in comparison with the

power supplied to the heater. The power consumed by the fan was 2.4 W while the

power consumption of the heater was on the order of 0.5 W. The power consumed by the

fan was included in the average and maximum power calculations for this reason.

4.5.1 Microheater testing results

The first case examined was for the heating and cooling rates of the microheater

system with the cooling fan and no heat sink. The cooling fan was only turned on during

cooling from 94 to 54 °C. The microheater was suspended in air for this test. There was

113

no test section on the heater. The purpose of this test was to determine the performance

of the heater and control system only. The results of the test are shown in Figure 4.27.

45

50

55

60

65

70

75

80

85

90

95

100

0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)

Tem

pera

ture

(°C

)

MicroheaterSetpoint

Figure 4.27 Heating and cooling characteristics of microheater with cooling fan, no heat

sink, 12 V input

The maximum heating rate for this configuration was 16.7 K/s while the maximum

cooling rate was 7.8 K/s. The average cooling rate was 4.0 K/s. The average heating

rates were 9.2 K/s and 3.6 K/s for the intervals between 54 and 72 °C and between 72 and

94 °C respectively. The average heating rate between 72 and 94 °C was low considering

that the maximum heating rate during this interval was 16.7 K/s. This can be attributed to

poor control of the microheater during this interval and is not a property of the heater.

The case with the cooling fan and the microheater mounted on top of an

aluminum heat sink was examined. The configuration of the setup is shown in Figures

114

4.25 and 4.26. The cooling fan was turned on only during cooling. The results of that

testing are shown in Figure 4.28.

45

50

55

60

65

70

75

80

85

90

95

100

0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)

Tem

pera

ture

(°C

)

MicroheaterSetpoint

Figure 4.28 Heating and cooling characteristics of microheater with cooling fan, heat

sink, 12 V input

The maximum cooling rate for this scenario was 14.6 K/s while the maximum heating

rate was 13.4 K/s. The average cooling rate between 94 and 54 °C was 8 K/s. The

average heating rate between 54 and 72 °C was 9 K/s while the average heating rate

between 72 and 94 °C was 11 K/s.

The power requirements of the microheater and fan were recorded. A summary

of the results along with the heating characteristics of the microheater are shown in Table

4.3. The maximum power required by the microheater system with and without the heat

sink was 2.4 W which is the power required to run the fan during cooling. During

cooling the heater is not on. The average power over a cycle for the microheater without

115

heat sink was 0.5 W. The average power over a cycle for the case with the heat sink was

1.0 W.

Table 4.3 Comparison of microheater operation with and without heat sink

Max Heating

rate (K/s)

Max Average heating

rate (K/s)

Max Cooling

rate (K/s)

Average cooling

rate (K/s)

Max Power consumption

(W)

Average power

consumption over a cycle

(W)Microheater without heat sink 16.7 9.2 7.8 4.0 2.4 0.5Microheater with heat sink 13.4 11.0 14.6 8.0 2.4 1.0

The purpose of the heat sink was to increase the cooling rate of the microheater. The

addition of the heat sink doubled the average cooling rate and increased the maximum

cooling rate by 7.8 K/s. The heat sink reduced the maximum heating rate, however,

because some of the heat energy was traveling to the heat sink. The average heating rate

actually increased with the heat sink, however, because in the case without heat sink there

was poor control at the 94 °C setpoint. The cost of using the heat sink can be seen in the

average power consumption column, however. The microheater needed twice as much

power on average to maintain the required temperatures.

116

5. MICROCHAMBERS FOR PCR

Silicon microchambers are one option for thermal cycling on a microheater or on

another flat, heated surface. Daniel et al. designed a silicon microchamber for rapid

thermal cycling. The researchers recognized that in order to perform rapid PCR the

thermal mass of the system must be small. A schematic of the device that was fabricated

is shown in Figure 5.1 (Daniel, Iqbal et al. 1998).

Figure 5.1 Silicon reaction chamber fabricated by Daniel et al.

The reaction chamber consists of a well fabricated from silicon that is suspended by four

beams. The reaction volume was designed to hold 2 µL of PCR reaction mixture along

with 1 µL of silicone oil, used to prevent evaporation.

117

Figure 5.2 Simplified fabrication sequence used by Daniel et al.

A simplified process sequence for the

device is shown in Figure 5.2 (Daniel, Iqbal et

al. 1998). The reaction vessel was fabricated

from a 400 µm thick silicon wafer with a silicon

nitride layer on both sides. Reactive ion etching

was next used to pattern the nitride layer on the

top surface to define the areas that would be

etched with KOH. The wafer was then etched

in KOH forming the pits shown in Figure 5.2.

The central pit is the reaction chamber while the

pits on each side are to provide thermal

isolation of the reaction volume. The two

lateral pits have a web of silicon nitride above

them. The purpose of the webbing was to prevent

oil from the reaction from creeping into the

thermal isolation pits. The idea behind this was that if the webbing was small enough, it

would prevent liquid from flowing into it because of surface tension. Platinum heaters

were patterned on the bottom of the reaction volume with a lift off technique using

chrome as an adhesion layer. The reaction chamber was coated with a SiO2 layer using

plasma enhanced chemical vapor deposition. The reaction volume then had a bovine

serum albumin treatment. It was found that this surface treatment was necessary to make

the microchamber PCR friendly by blocking adsorption of PCR reactants to the chamber

walls (Daniel, Iqbal et al. 1998).

118

This reaction volume was successfully used to amplify a 260 bp DNA segment.

The protocol was 5 seconds at 94 °C, 7 seconds at 55 °C, and 5 seconds at 72 °C. The

time constant for this system was 0.4 seconds (Daniel, Iqbal et al. 1998).

Shoffner et al. have also found it necessary to perform a surface treatment on

silicon chambers in order to perform PCR. The researchers first tested the effect of

treated powder silicon on PCR in standard polypropylene PCR tubes in commercially

available thermal cyclers. This was accomplished by incubating the silicon powder at

room temperature overnight in excess silanizing agent in a test tube. The excess

silanizing agent was then removed and the powder was dried overnight in an oven at 70

°C. The powder was then rinsed with distilled, deionized water and dried overnight in

the oven again. The powder was next separated into tubes and treated with 1 mL of 10

mg/mL polymer solution. The polymer solutions were prepared with 0.1 M Tris buffer.

The following polymers were used: poly-α-alanine, poly-L-aspartic acid, polyglycine,

poly-L-Leucine, poly-DL-phenylalanine, poly-DL-tryptophan, poly-L-lysine,

polyvinylpyrrolidone, polyadenylic acid, polymaleimide or maleimide. The silicon

powder was incubated at room temperature overnight at room temperature. The excess

solution was removed and the powder was oven dried overnight at 45 °C. It was found

that untreated silicon completely inhibited the reaction. Silicon powder treated with a

silicone fluid consisting mainly of dichlorooctamethyltetrasiloxane followed by a coating

of polyadenylic acid compared well with PCR run with no silicon powder (Shoffner,

Cheng et al. 1996).

This information was used to test the efficacy of PCR in silicon microchambers

(Shoffner, Cheng et al. 1996). The microchambers tested were 14 x 17 x 0.12 mm. The

119

reaction chambers had a Pyrex glass cover anodically bonded to them. The silicon chips

were first treated with the silanizing agent for 15 minutes at room temperature and then

removed. The chip was rinsed with distilled, deionized water and dried. The PCR chip

was filled with one of the various polymer solutions listed above and allowed to sit for 1

hour at room temperature. The polymer solution was removed using a vacuum, and the

chip was rinsed, and dried. SiO2 and Si3Ni4 layers of thickness 1000 Å were also applied

to some of the chips to determine the affect of these layers (Shoffner, Cheng et al. 1996).

The surface treatments enhanced PCR in the silicon microchambers compared

with untreated silicon (Shoffner, Cheng et al. 1996). The untreated silicon

microchambers showed some inhibition of the PCR reaction while those treated with the

silanizing agent and polymer coating showed results comparable to those performed in a

polypropylene tube in a commercially available thermocycler. The results from the

silicon chips treated with the silanizing agent and the polymer coating were not

repeatable, however, and this was attributed to a non-uniform surface treatment. Another

explanation was the coating may degrade over the course of the PCR reaction. SiO2 and

Si3N4 surfaces were also tested. These surfaces had a more uniform coating. The bare

silicon surface and silicon coated with Si3N4 showed inhibition of PCR. The SiO2 coated

chips showed the best performance as compared with the polypropylene tube in the

commercial thermal cycler. The results from these chips were repeatable. (Shoffner,

Cheng et al. 1996).

Silicon chambers have been shown to be PCR compatible when coated as

discussed above. Silicon microchambers are time consuming to fabricate, however.

KOH etching of silicon progresses at 1 µm/minute (Kovacs 1998). For a 500 µm wafer it

120

would take 7 hours to etch 400 µm of the material away. It has been shown above that

these wafers need additional surface treatment in order to be PCR compatible.

Researchers have focused their attention on building PCR reaction chambers out of

materials such as polycarbonate and polypropylene for these reasons.

Yang et al. have successfully completed PCR in a polycarbonate reaction volume.

The device fabricated consisted of three layers of polycarbonate thermally bonded. Each

polycarbonate layer was 250 µm thick. The top and bottom of the chamber were made of

clear polycarbonate while the middle section was made of black polycarbonate. The

polycarbonate was cut using a C02 laser engraving system. The chamber itself was a

serpentine channel, designed to decrease dead volume and bubble trapping. A picture of

the chamber is shown in Figure 5.3 (Yang, Liu et al. 2002).

Figure 5.3 Polycarbonate chamber fabricated by Yang et al.

The channel is 1.5 mm wide, 0.25 mm high, and can hold 40 µL of PCR reaction fluid.

The holes at the end of each channel were 0.7 mm in diameter. The chambers were

bonded using a Carver Inc. thermal press with a platen temperature of 133 °C and a

pressure of 1400 psi for 2 hours. The inlet and outlet holes were sealed with double sided

adhesive tape after the PCR solution was loaded. A layer of paraffin was then put on top

of the tape followed by a polycarbonate plate. A hand press was used to press down on

121

this assembly, causing the tape to be pressed down into the filling holes, creating a strong

seal (Yang, Liu et al. 2002).

This reaction chamber was used to complete a rapid PCR reaction (Yang, Liu et

al. 2002). The chamber was cycled using a dual Peltier chip setup. The chamber was

sandwiched in between two thermoelectric coolers. PCR was completed using this setup

with a total reaction time of 30 minutes. 15 seconds was spent at each temperature

plateau in the PCR cycle. The PCR yield was comparable with that achieved with a

conventional thermal cycler (Yang, Liu et al. 2002).

The PCR reaction mixture used by these researchers was modified in order to

accommodate the large surface area to volume ratio of the chamber. Polyethylene glycol

was added to the PCR reaction buffer to increase the yield. This approach has been

shown to be effective for increasing the PCR yield in polyimide microchips (Giordano,

Ferrance et al. 2001). The PCR yield was highest when using 0.75 % polyethylene glycol

8000. Different molecular weights of polyethylene glycol are available. The 8000 refers

to the molecular weight. Bovine serum albumin was also added to the reaction volume at

a concentration of 250 µg/mL to increase the yield. It was thought that the combination

of these two treatments helped block adsorption of any of the PCR reactants to the

chamber walls (Yang, Liu et al. 2002).

Other researchers have also produced polycarbonate reaction chambers. Chen et

al. has designed a polycarbonate reaction vessel for electrokinetic flow. Three

temperature zones were maintained and the fluid was moved from temperature zone to

temperature zone as required (Chen, Musundi et al. 2005). The channel designed had a

width of 100 µm, a depth of 70 µm, and was 7.9 cm long. The volume of the reaction

122

vessel was 0.55 µL. The chamber was fabricated using a brass molding die and hot

embossing in polycarbonate. The brass mold insert was heated to 180 °C and pressed

into the polycarbonate sheet at 1000 lbf for 4.5 minutes. The cover for the chamber was

thermally bonded in an oven at 150 °C for 20 minutes. The PCR reaction was degassed

in a vacuum after loading into the chamber to reduce the formation of air bubbles. 2 %

ethylene glycol was also added into the PCR reaction mixture to increase the boiling

point of the mixture and to reduce bubble formation in the device (Chen, Musundi et al.

2005).

Bubble formation in polycarbonate chambers can be a problem as noted by the

efforts taken by Chen et al. to reduce bubble formation in the chamber. Zhao et al. also

noticed problems caused by bubbles in a silicon microchamber. It was recommended that

the reaction volume not be filled completely. A small bubble should be left in the

reaction chamber to allow for expansion of the liquid upon heating. The researchers

found that the bubble should be located at the edge of the reaction vessel. Bubbles that

arose in the middle of the reaction vessel expanded upon heating and forced liquid to the

edges of the reaction chamber. This caused a problem with temperature uniformity in the

device because the temperature sensor was located directly below the heating chamber.

If all of the fluid was pushed out into the filling capillaries, that reaction volume would

not reach the required temperature (Zhao, Cui et al. 2003).

It was necessary to use a surface treatment or to add substances to the PCR

reaction vessel to inhibit adsorption of protein to the chamber walls in the research

described above that used polycarbonate and silicon. Surface treatments were done to

silicon that made the surface hydrophilic. Polycarbonate can also be made hydrophilic.

123

An oxygen plasma etch on a polycarbonate sample of thickness 1.2 mm was shown to

dramatically decrease the water contact angle (Larsson and Derand 2001). The

researchers used a Plasma Science PSO500 reactor. A 100 standard cubic centimeters

per minute (sccm) oxygen flow rate was used with the best results coming with an RF

power of 300 W. The water contact angle after this 5 minute plasma treatment was 3 °

compared to contact angles around 70.1 ° for untreated polycarbonate. The samples were

stored in aluminum foil to test the stability of the plasma treatment in dry storage.

Samples still had contact angles around 20 ° after six months of storage wrapped in

aluminum foil. This behavior was observed for polycarbonate treated with reactive ion

etch (RIE) bias voltages of 480-600 V. Intense plasma treatments also produced changes

in the Young’s modulus of the material of up to 1 GPa (Larsson and Derand 2001).

Polycarbonate can also be treated with UV light to increase the surface

hydrophilicity (Liu, Ganser et al. 2001). A 4 W manual UV lamp was used for the

surface treatment of the polycarbonate. The polycarbonate that was treated was 1 mm

thick. The polycarbonate was exposed to UV light for variable amounts of time ranging

from 0 to 13 hr. As time increases the surface becomes increasingly hydrophilic until the

contact angle levels off after 5 hours at 22 °. The polycarbonate was able to be thermally

bonded after surface treatment and no detectable changes in mechanical properties were

present (Liu, Ganser et al. 2001).

124

5.1 Microchamber design considerations

The purpose of the reaction chamber is to distribute the reaction volume in a thin

layer and have a thin layer separating it from the microheater to ensure rapid heating. A

2-D finite element heat transfer package was used to develop a sense of what chambers

might be acceptable. The first material to be examined was polycarbonate.

Polycarbonate can be thermally bonded (Bilitewski, Genrich et al. 2003; Ye, Yin et al.

2005), is easy to cut, and can be commonly found in thin sheets of 250 µm. The

microheater that was available was 5 mm square. A chamber 5 mm square or smaller

was desired for this reason. A reaction volume 5 mm square with a depth of 250 µm

would hold 6.25 µL of PCR solution. The 2-D model that was used to model this

situation is shown in Figure 5.4. The water was assumed to be static in this calculation.

Figure 5.4 2-D heat transfer model for polycarbonate chamber. All dimensions in µm.

The boundary condition around the outside of the polycarbonate, glass, and heat sink was

free convection with a value of 5 W/m2 K. A temperature boundary condition was set on

the 5 mm long border between the glass and the polycarbonate, directly beneath the water

section, to simulate the surface of the microheater. The results of running the model with

an initial system temperature of 72 °C are shown in Figure 5.5.

125

50

55

60

65

70

75

80

85

90

95

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Time (seconds)

Tem

pera

ture

(°C

)

Upper cornerMiddle top

Figure 5.5 2-D conduction model of 5 mm square polycarbonate chamber on 5 mm

square microheater

The lower trace is for a node in the upper left hand corner of the chamber while the upper

trace is for a node in the top center of the chamber. Both of these nodes were in contact

with the water. The time constant for the top middle node was 2.5 seconds while the time

constant for the upper corner node was 4 seconds. Considering that it takes four time

constants for the reaction volume to reach 98 % of its final value, this may not be an

optimal design for rapid PCR.

The reaction volume for this setup was reduced to see the effect of a smaller

reaction volume on heating between 72 and 94 °C. The reaction chamber in this second

model was 3 mm square. The polycarbonate surrounding the reaction chamber had the

same dimensions as shown in Figure 5.4. This meant that for the 5 mm square heater 1

126

mm of heater lay outside of the perimeter of the PCR solution on each side. The volume

of this reaction chamber is 2.2 µL. The results of the FEA model are shown in Figure

5.6.

50

55

60

65

70

75

80

85

90

95

100

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20Time (seconds)

Tem

pera

ture

(°C

)


Figure 5.6 2-D conduction model of 3 mm square polycarbonate chamber on 5 mm

square microheater

The upper corner trace corresponds to a node at the upper corner of the reaction vessel in

contact with the PCR mixture, where the temperature was expected to increase slowest.

The middle top trace represents a node at the top center of the PCR reaction vessel in

contact with the solution. Nodes at the top surface of the liquid were of interest because

these points would be the slowest to respond. The temperature uniformity in the reaction

vessel is much better in this case than in the case with the 5 mm reaction chamber. The

time constant for the upper corner node was 2.5 seconds while the time constant for the

127

middle top node was 2 seconds. The time constant for the top middle node remained the

same while the time constant for the upper corner node was decreased by 2 seconds

compared to the 5 mm reaction chamber. The upper corner node showed a decrease in

time constant because of the overhang of the heater in this case.

A 3 mm square reaction chamber with the same thickness shown in Figure 5.4

made from silicon with a 5 mm square heater centered beneath it was also examined to

see the effect of a different material on the time response of the chamber. Silicon has a

thermal diffusivity of 89 x 10-6 m2/s while the thermal diffusivity of polycarbonate is 138

x 10-9 m2/s(Incropera and DeWitt 1996). The silicon chamber will heat up much faster

than the polycarbonate one. The result of that calculation is shown in Figure 5.7.

50

55

60

65

70

75

80

85

90

95

100

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1Time (seconds)

Tem

pera

ture

(°C

)


Figure 5.7 2-D conduction model of 3 mm square silicon chamber on 5 mm square

microheater

128

Comparing Figures 5.6 and 5.7, it is clear the dramatic affect changing the reaction

chamber vessel material has on the speed of heating. The time scale on Figure 5.7 is

different because of the very rapid heating that the model predicts. The time constant for

a node at the middle top of the reaction volume in contact with the liquid was 0.01

seconds while the node at the upper corner of the reaction mixture in contact with the

liquid had a time constant of 0.03 seconds. This is very small compared to the 2 second

time constant of the polycarbonate chamber of identical dimensions.

The same chamber was analyzed to determine the speed at which a glass

microchamber could be heated on the 5 mm square microheater. The thermal diffusivity

of glass is 328 x 10-9 m2/s while the thermal diffusivity of polycarbonate is 138 x 10-9

m2/s. This means that the glass will heat up faster than polycarbonate. The analysis

indicated that the time constant for a node at the upper corner of the reaction chamber

was 1.1 seconds while a node at the top middle of the reaction mixture had a time

constant of 0.9 seconds indicating a uniform reaction chamber temperature.

Glass capillaries have been used for PCR in convection thermocyclers (Loeffler,

Henke et al. 2000). Glass capillaries have also been coated with indium tin oxide which

served as both a heater and a temperature sensor. PCR was performed in ten minutes

using this device (Meldrum, Evensen et al. 2000). A heat transfer model was made of a

glass capillary on the microheater to determine the time constant of the system. A

schematic of the model is shown in Figure 5.8.

129

Figure 5.8 Schematic of 800 µm capillary on microheater

The capillary outer diameter was 800 µm with an inner diameter of 400 µm. The

boundary condition around the outside of the capillary was assumed to be 5 W/m2K

except for at the bottom of the capillary where a 200 µm segment had a temperature

boundary condition. The time constant for heating between 72 and 94 °C was 1.5

seconds for a node in the water area farthest from the heater. The time constant for this

system lies in between the 3 mm square glass chamber and the polycarbonate chamber.

The biggest difference between these chambers is in the cross section of the chamber.

The glass capillary has a reaction chamber with a 0.13 mm2 cross section area while the 3

mm square chamber has a cross sectional area of 0.75 mm2. Even though the glass

capillary reaction chamber has a smaller cross sectional area than the planar design, the

time constants for both systems are comparable because planar design utilizes the heater

surface better and minimizes thermal penetration depth.

130

The above analysis demonstrates that reaction volumes made of different

materials with different sizes have different thermal time constants. The trend shows that

smaller reaction volumes and smaller penetration depths aid in heating speed. Materials

with higher thermal conductivities, such as silicon, perform much better than those with

low thermal conductivity materials like polycarbonate and glass.

A complete PCR reaction consists of many thermal cycles with temperatures

reaching as high as 94 °C. Liquid and air both expand, so it is necessary to know by how

much they expand so the reaction volume may be filled appropriately and to understand

how they should be sealed. PCR microchambers can be sealed with a drop of mineral or

silicone oil or by bonding material over the top of the filling hole (Nagai, Murakami et al.

2000; Matsubara, Kerman et al. 2005). One possible polycarbonate reaction volume is


Figure 5.9 Polycarbonate reaction vessel thermally bonded from three 250 µm thick

sheets

Water

Air bubble

131

A reaction vessel of this type is made out of three layers of polycarbonate thermally

bonded. The top layer has the filling holes, the middle layer contains the reaction vessel,

and the bottom layer has no features and is meant to seal the reaction vessel. A reaction

vessel of this type can be sealed with oil in the filling reservoirs or by bonding additional

pieces of polycarbonate over the filling holes after filling. If the ends are sealed through

polycarbonate bonding, the reaction vessel is closed and pressure may increase in the

volume during heating.

The central reaction area of a polycarbonate chamber sealed at the ends of the

type shown in Figure 5.9 was modeled. The model assumed that the system was closed

and rigid, that the air in the reaction chamber is an ideal gas mixture of air saturated with

water vapor, and that no new air bubbles arise in the reaction mixture. The governing

equations of this model are given in Table 5.1 along with a description of the model.

Table 5.1 Model to predict pressure in sealed two-phase system

Physical Meaning Equation (written as EES code) Equation Number

Total volume sum of vapor and liquid lvtot VVV += 5.1

Air component is ideal gas TRVP

mair

vairair = 5.2

Vapor pressure of water )1,,( == xTWaterpressureP ambw 5.3

Water vapor is ideal gas TRVP

mwater

vwvwv = 5.4

Liquid volume )0,,( ===

xTTWatervolumeV

mamb

lwl 5.5

Mass of water wwvw mmm += 5.6 Total pressure sum of partial pressures airw PPP += 5.7

132

where V refers to volume, P is pressure, R is the universal gas constant divided by the

molecular weight of air or water as specified by the subscript, m is the mass, T is

temperature, and x is vapor quality. The subscripts are defined as follows: w refers to

water, wl refers to liquid water, wv is the water vapor, i is an initial value, air refers to the

air-water vapor mixture in the chamber, and tot is a total amount. Pressure and volume

are EES functions that return the pressure and specific volume of water at the indicated

temperatures and qualities.

The model was run for temperatures between ambient and 100 °C. The volumes

used correspond to a reaction vessel measuring 5 mm square with a height of 250 µm.

The reaction vessel was filled with 4.2 µL of water with a 2 µL air bubble. The initial

pressure inside the reaction vessel was 1 atm. Pressure increased in the reaction chamber

as shown in Figure 5.10.

100

125

150

175

200

225

250

20 30 40 50 60 70 80 90 100

Temperature (°C)

Pres

sure

(kPa

)

Figure 5.10 Predicted pressure rise inside closed reaction vessel with water and 33% of

the volume filled with air

133

The predicted pressure at 95 °C is 210 kPa, or a gauge pressure of 110 kPa. A typical

cyanoacrylate glue (often called super glue) has a polycarbonate bonding strength of

5000 kPa (Loctite 2000). A gauge pressure of 110 kPa would create a 2.2 N force on a

circular filling hole of diameter 2 mm. Assuming that the cyanoacrylate is applied to an

area of 9 mm2, a pressure of 244 kPa would be present in the glued region. This value is

well below the polycarbonate bonding strength for cyanoacrylate making it feasible to

use cyanoacrylate to bond the cover when considering pressure only.

This model also shows that an air bubble present in the reaction vessel should also

shrink as a result of the expanding water. A graph depicting this behavior is shown in

Figure 5.11.

0.00.10.20.30.40.50.60.70.80.91.01.11.21.31.41.51.61.71.81.92.0

20 30 40 50 60 70 80 90 100

Temperature (°C)

Bub

ble

volu

me

(mm

^3)

Figure 5.11 Predicted bubble volume inside closed reaction vessel with water and 33%

of the volume filled with air

134

5.2 Microchamber fabrication equipment and techniques

Thermally bonded polycarbonate chambers of the type shown in Figure 5.12 were

fabricated. 250 µm or 125 µm thick polycarbonate film was used. The chamber profiles

were drawn in Corel Draw and then cut into the polycarbonate using a PII-30 vinyl cutter

from a company called GCC. The cutter was unable to cut all the way through the

polycarbonate so it was necessary to complete the cutting with a razor blade. The

individual polycarbonate pieces were then rinsed with isopropanol and thermally bonded

in an oven. The polycarbonate was sandwiched between two slabs of metal coated with

Teflon. The assembly was clamped together using a spring loaded clip. The assembly

was placed in the oven at 163 °C for 15 minutes and then removed. The chamber was

removed from the clamp four minutes later. The chambers were sealed with acetone,

methyl ethyl ketone, methylene chloride, or cyanoacrylate and polycarbonate.

Figure 5.12 Polycarbonate chamber fabricated from three 250 µm thick sections. All

dimension in mm

135

This particular chamber has a 5 mm square reaction chamber with 1 mm wide filling

capillaries. The reaction volume of this chamber is 6.2 µL. Polycarbonate reaction

vessels of this type were also fabricated with a 3 mm square reaction chamber. These

chambers had a reaction volume of 2.2 µL.

Chambers with an alternative design were made by drilling into a 2 mm thick

sheet of Hyzod polycarbonate from Sheffield Plastics, Inc. A Dremel Multipro model

395 was mounted in a Craftsman model 572.530320 drill press stand. A 3 mm diameter

router bit was used to create the profiles. Some chambers were sealed by placing a drop

of mineral oil on top of them. The mineral oil used was supplied by Biomerieux, part

number 70100. A schematic of one of these chambers is shown in Figure 5.13.

Figure 5.13 Countersunk polycarbonate chambers created with router bit

136

The polycarbonate chambers shown above were designed to put the PCR reaction volume

in the lower section. The thin layer surrounding the reaction volume was intended as an

oil layer. The purpose of the oil was to prevent solution evaporation.

Polycarbonate chambers of the type shown in Figure 5.13 were made hydrophilic

using an oxygen plasma etch in a Trion Technology minilock reactive ion etcher. The

polycarbonate was first rinsed with isopropanol and deionized water. The polycarbonate

pieces were than placed on a Thermolyne type 1900 hot plate with Taylor temperature

controller, model 9940 at 110 °C for ten minutes. The RIE was set to 125 W with a

pressure of 250 mTorr and an oxygen flow rate of 100 sccm. The oil reservoir was made

hydrophobic after RIE treatment by wiping it with isopropanol.

Polycarbonate was also made hydrophilic by using a portable UV light source.

The light source was a Spectroline UV lamp, model UV-4BSW from Spectronics

Corporation. The bulb used in this device emitted UV light at 254 nm.

Glass capillaries were also used as a reaction volume that had the potential for

rapid thermal cycling. The glass capillaries had an outer diameter of 890 µm with an

inside diameter of 400 µm. The capillaries were supplied by Drummond Scientific Co.,

Broomall PA.

5.3 Microchamber testing

Polycarbonate chambers were the main focus of chamber manufacture. Two

types of polycarbonate chambers were tested. The first type was the countersunk

chamber discussed earlier. The second type was the thermally bonded polycarbonate

chamber. Different methods were used to attempt to seal the thermally bonded

137

polycarbonate. Glass capillaries were also studied to determine their effectiveness for

rapid PCR thermocycling.

5.3.1 Countersunk polycarbonate chambers

Countersunk polycarbonate chambers were tested. The testing of these chambers

was done on the thermoelectric cooler (TEC) with an oil layer on top of the water with

blue food coloring. The testing involved a simulated PCR cycle consisting of an initial

six minute heating period at 95 °C and 30 cycles of PCR with 60 seconds at each

temperature and a final elongation step of 5 minutes at 72 °C. Untreated polycarbonate

chambers were tested. The chambers were simply drilled and then tested. Untreated

drilled polycarbonate wells exhibited total evaporation of the water. Large air bubbles

formed within the liquid in the chambers and eventually, all of the reaction mixture

would evaporate. A picture of these bubbles is shown in Figure 5.14.

Figure 5.14 Bubble formation in untreated polycarbonate chambers covered with

mineral oil

138

The bubbles would first start to form in independent locations and then coalesce into one

large bubble. These bubbles seemed to displace the oil allowing the water to evaporate.

The bubbles were assumed to be from trapped air on the polycarbonate surface. It

is noted that glass vessels were tested (larger geometry) similarly and no air bubbles were

observed. One of the differences between glass and polycarbonate is that glass is

hydrophilic while polycarbonate is hydrophobic. Attempts were made to make the

polycarbonate chambers hydrophilic for this reason. The first attempts at making

polycarbonate hydrophilic were with an O2 plasma etch. Different reactive ion etch

powers were used to determine the optimal power. A picture showing the results of these

tests is shown in Figure 5.15.

Figure 5.15 2 mm thick polycarbontate after O2 plasma etch at A)100 W, B) 200 W, C)

250 W

The 100 W sample showed a decreased contact angle versus untreated polycarbonate

with no deformation of the sample. The water on the surface of the 200 W sample was

almost flat to the surface as shown in Figure 5.15. The 200 W case had slight

deformation. The 250 W case showed significant deformation. Further testing showed

that the optimal level for O2 plasma treatment of a 2 mm thick polycarbonate sample

using the RIE mentioned in the fabrication equipment section was 125 W based on

physical distortion and water contact angle.

A B C

139

O2 plasma etching was done on the countersunk polycarbonate chambers. After

treatment the oil layer reservoir surface was wiped with isopropanol to remove the

surface charge on that area. This led to a hydrophilic reaction volume and a hydrophobic

oil reservoir. The purpose of this was to isolate the water in the bottom of the chamber.

These chambers allowed a simulated PCR cycle to be run in the chambers without any

observed evaporation. Bubbles still appeared in the solution, but they did not coalesce

nor disturb the oil layer as they did in the untreated polycarbonate case.

5.3.2 Glass capillaries

Glass capillaries were used frequently for PCR in this effort because it was found

that glass gave more consistent results than polycarbonate. Glass capillaries with a 400

µm inside diameter and an outside diameter of 890 µm proved to be very effective

vessels for PCR. They could easily be filled with PCR solution and capped with oil

through capillary action. There were never any problems with evaporation with these

chambers when heating on the TEC or microheater. The time constant of glass

capillaries filled with ethylene glycol was similar to those predicted by the heat transfer

model described in Section 5.1. Ethylene glycol was used for this test because it has a

boiling point of 197 °C. This allowed the capillary to be filled without having to seal the

ends with oil. The thermal diffusivity of ethylene glycol is 93 x 10-9 m2/s compared with

145 x 10-9 m2/s for water. This means that ethylene glycol will respond to changes in

temperature slower than water. Heat transfer analysis showed that capillaries filled with

water and capillaries filled with ethylene glycol should perform similarly. A graph

showing the performance of a glass capillary on the microheater is shown in Figure 5.16.

140

253035404550556065707580859095

100

0 5 10 15 20 25 30 35 40 45 50 55 60 65

Time (seconds)

Tem

pera

ture

(°C

)Heater temperatureInside capillary temperature

Figure 5.16 PCR cycle showing 5 mm square microfabricated heater and temperature

inside a 400 µm inside diameter glass capillary filled with ethylene glycol

A thermocouple was inserted into the glass capillary filled with ethylene glycol to obtain

the inside capillary temperature. The capillary temperature matches the heater

temperature very closely during cooling. During heating from 72 to 95 °C the time

constant of the heater is 1 second while the time constant for the capillary was 3 seconds.

This is similar to the calculated time constant of 1.5 seconds obtained in the 2-D analysis

which was obtained with a temperature step input.

5.3.3 Thermally bonded polycarbonate chambers

Thermally bonded polycarbonate chambers of the type shown in Figure 5.12

consisting of three 250 µm thick layers were tested on the microheater. The first

141

chamber tested was a chamber that had a 5 x 5 mm reaction chamber. The reaction

chamber volume was 6.2 µL. The chamber was filled with ethylene glycol and the

capillary channel was cut off at a point just before the square reaction chamber so a

thermocouple could easily be inserted into the chamber. A clamp on the heater setup

held the test section securely in place. The clamp was made of polycarbonate and was

off of the heated portion of the chamber. Grease was placed in between the heater and

the polycarbonate surface to enhance heat transfer. The results of that test are shown in

Figure 5.17.

253035404550556065707580859095

100

0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)

Tem

pera

ture

(°C

)

Heater temperature

Inside chambertemperature

Figure 5.17 PCR cycle showing 5 mm square microfabricated heater and temperature

inside a 5 mm square polycarbonate reaction volume filled with ethylene glycol

When the heater temperature was at 72 °C, the temperature inside the chamber stabilized

at 68 °C as shown in Figure 5.17. This is indicative of a thermal resistance between the

heater and the thermocouple. The time constant for heating between the 72 and 94 °C

setpoint for the reaction chamber was ten seconds, which is much larger than the 2.5 – 4

142

seconds calculated with the FEA model. This poor performance is attributed to poor

thermal contact between the chamber surface and the heater.

A chamber was tested that was made by thermally bonding three 250 µm thick

pieces of polycarbonate. This chamber only had a central reaction chamber of 3 x 3 mm,

however. The volume of the central reaction chamber was 2.2 µL. This chamber was

placed onto the microheater in the exact same way as the 5 x 5 mm chamber. The

temperature inside the chamber was recorded and is shown in Figure 5.18.

253035404550556065707580859095

100

0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)

Tem

pera

ture

(°C

)

Heater temperature



temperature inside a 3 mm square polycarbonate reaction volume filled with ethylene

glycol

The 3 x 3 mm polycarbonate chamber followed the heater temperature closer than the 5 x

5 mm chamber. The time constant for heating between 72 and 94 °C for the 3 x 3 mm

chamber was 3 seconds. The time constant of the heater itself was 1.5 seconds as shown

143

in Figure 5.18. This value is comparable with the 2 second time constant calculated with

the conduction analysis done with a step temperature input.

A thermally bonded microchamber was tested that had reaction chamber 3 mm

square with a 125 µm thick polycarbonate section bonded to the top and bottom of the

250 µm thick middle section. The 125 µm thick sections were in place of the 250 µm

thick sections that were bonded to the top and bottom in prior experiments. The purpose

of the experiment was to see if having a thinner polycarbonate section on the bottom of

the chamber would increase the heat transfer to the chamber. The chamber was once

again filled with ethylene glycol and a thermocouple was inserted into the reaction

chamber. The chamber was clamped to the microheater. The results of this test are

shown in Figure 5.19. The time constant for the chamber during heating from 72 to 94

°C was 3 seconds while the time constant for the heater was 1.2 seconds. This value is

equal to the time constant of the same chamber with a 250 µm thick top and bottom

section.

144

253035404550556065707580859095

100

0 5 10 15 20 25 30 35 40 45 50 55 60Time (seconds)

Tem

pera

ture

(°C

)Heater temperature



temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick

top and bottom section filled with ethylene glycol

A 3 x 3 mm reaction chamber 250 µm thick with a 250 µm thick polycarbonate

section bonded to the top and a 125 µm thick section bonded to the bottom was tested to

determine if this variation would perform better than the case with a 125 µm thick sheet

of polycarbonate bonded to the top and bottom. The performance of this chamber is


145

253035404550556065707580859095

100

0 5 10 15 20 25 30 35 40 45 50 55 60

Time (seconds)

Tem

pera

ture

(°C

)

Heater temperature



temperature inside a 3 mm square polycarbonate reaction volume with a 125 µm thick

bottom section and a 250 µm thick top section filled with ethylene glycol

The time constant for the reaction chamber during heating between 72 and 94 °C for this

case was 2 seconds. The performance of this chamber is superior to any of the chambers

tested above. This is clearly seen by comparing the graphs from all of the above analysis

and noting the correlation between the heater temperature and the temperature inside the

chamber. The microheater showed poor control at the 72 °C setpoint in this case.

5.3.3.1 Thermally bonded polycarbonate chamber sealing

A thermally bonded polycarbonate chamber with a 3 x 3 mm reaction chamber

with no sealing on the ends was tested on the microheater to assess its behavior. This

146

chamber was filled and placed on the microheater at 95 °C. The fluid was immediately

pushed out of the central chamber and out to the filling holes.

A thermally bonded polycarbonate chamber with a 3 x 3 mm reaction chamber

was sealed with polycarbonate and cyanoacrylate. The surface of the chamber was first

cleaned with isopropanol after filling with a 5 % ethylene glycol and water solution dyed

blue. Two small pieces of 250 µm thick polycarbonate were cut and bonded to the

chamber using cyanoacrylate. The assembly was then placed into a clamp for 5 minutes

to strengthen the bond. The chamber was then placed onto the microheater at 95 °C for

600 seconds, 30 PCR temperature cycles with 20 seconds at each temperature, and 300

seconds at 72 °C. The fluid in the central reaction volume was not immediately expelled.

An air bubble did form in the middle of the reaction volume however, and did grow. The

air bubble initially in the chamber occupied 20 % of the volume and was located on the

edge of the chamber. The filling capillaries were completely full for this test. The

bubble appeared to grow over the course of the test and upon completion occupied 30 %

of the volume and was centrally located. This was problematic because the goal is to

keep the PCR solution in the PCR chamber

One of the 3 x 3 mm square polycarbonate reaction chambers was inserted into a

vacuum chamber for ten hours prior to filling to assess the effect on bubble formation in

the chamber. The chamber was filled with water and sealed with cyanoacrylate. A small

air bubble existed in the reaction chamber prior to thermal cycling. The capillaries

leading up to the reaction chamber were filled only halfway. The fluid in the reaction

chamber was immediately pushed out into the filling capillaries upon placing on the

microheater. The difference between this test and the prior one is that the capillaries in

147

this case were only partially filled while the capillaries in the above case were completely

full. This indicates that it is important for this configuration to have the filling capillaries

completely full.

Another polycarbonate chamber that was not placed into the vacuum chamber

was filled with water. The reaction chamber was filled such that only a small bubble

existed in the reaction chamber. The filling capillaries were completely filled with water.

The filling holes were then sealed with polycarbonate/cyanoacrylate and the chamber was

placed onto the microheater. The fluid in the reaction chamber was not immediately

expelled. The chamber was heated at 95 °C for 10 minutes, cycled with 20 second time

intervals 30 times, and then heated at 72 °C for 5 minutes. The fluid volume in the

reaction chamber decreased during heating at 95 °C while the vapor bubble expanded.

During cooling the opposite happened. The volume of fluid in the central reaction

chamber appeared to be the same as before thermal cycling after the test. This test

showed that a completely full polycarbonate chamber with water could be sealed with

polycarbonate and cyanoacrylate.

A polycarbonate chamber of the same type was filled with ethylene glycol to

determine the mechanism of bubble formation. Ethylene glycol has a much higher

boiling point then water. If the bubbles in the reaction chamber were due to air trapped

between the fluid and the polycarbonate, then they should also appear in the ethylene

glycol. The chamber was placed onto the microheater at 95 °C. The air bubbles in the

reaction chamber did appear.

The results of these tests indicate that trapped air on the polycarbonate surface is

liberated during heating. The air forms bubbles that tend to push the fluid in the reaction

148

chamber. These air bubbles exert a pressure in addition to the pressure created by the air

bubble already present in the chamber and the expansion of the fluid. This can be

detrimental because the goal is to keep the reaction volume in a stationary location. The

best case scenario was for a polycarbonate chamber that had the filling capillaries

completely full and a small bubble in the reaction chamber. Glass capillaries do not

exhibit this behavior perhaps because of their hydrophilic nature. The water is attracted

to the surface and hence expels the air from the capillary upon filling. Fluid placed in

them will stay in them during heating as long as they are sealed with mineral oil.

149

6. SUMMARY, CONCLUSIONS, AND FUTURE WORK

6.1 Summary

The primary objective of this work was to investigate alternative system designs

to achieve high speed PCR. A secondary objective was to minimize power consumption.

Each of the design alternatives tested had a unique set of features that may be useful for a

particular application. Since the focus was on speed and power consumption, much of

the description focuses on these parameters. A summary comparison of all of the designs

is given in Table 6.1. The microheater was found to yield the fastest cycling with the

lowest power consumption, but due to a series of problems with vessel design, definitive

PCR results were not obtained in this configuration.

Table 6.1 Comparison of thermocyclers

Heater configurationMax heating rate

(K/s)

Average heating rate

(K/s)

Max cooling

rate (K/s)

Average cooling rate

(K/s)

Max power consumption

(W)

Average power over a cycle

(W)TEC with cooling fan 5.2 1.8 6.6 5.8 23.7 11.3TEC without cooling fan 5.8 2 5.2 4.3 23.7 7.9Techne themocycler 3.6 2.8 2.0 1.4 264 145Single fan convective thermocycler 2.0 1.2 1.8 1.2 604 255Double fan convective thermocycler 4.2 2.6 7.9 5.3 1371 645Microheater without heat sink 16.7 9.2 7.8 4.0 2.4 0.5Microheater with heat sink 13.4 11.0 14.6 8.0 2.4 1.0

A listing of the major tasks and accomplishments of the work is given in Sections

6.1.1-6.1.3.

150

6.1.1 Testing of Peltier thermocyclers

• An independent thermoelectric cooler was tested. Two configurations of this

device were tested. The two configurations tested differed in whether the fan on

the heat sink was on or off.

• The heat flux produced by the TEC was modeled using the lumped capacitance

method. The model was verified by putting the heat flux results into a 2-D heat

conduction program.

• A commercially available Peltier thermocycler was tested to assess the heating

and cooling rates and power consumption.

• The commercial Peltier device was used to study nucleotide insertion rate and

PCR in glass capillaries.

6.1.2 Testing of convective thermocyclers

• Two different convective thermocyclers were tested. One of the convective

thermocyclers tested had a single fan. The second convective thermocycler tested

had two fans. One of the fans was responsible for flowing air over the heated coil

in heating mode. The second fan was used to introduce a cool, high velocity air

stream into the vent tube.

• Placing mixing walls into the vent tube and wrapping the outside with insulation

provided the most uniform temperature within the vent tube.

• The air velocity in the convective thermocyclers was determined using a control

volume approach.

151

• The lumped capacitance method and a 2-D heat transfer package were used to

model the behavior of a polypropylene PCR reaction tube with 10 µL of water in

it in the convective thermocycler.

• The lumped capacitance model was used to investigate the behavior of reaction

vessels of different geometries and materials in the convective thermal cycler.

The effect of different air velocities was also examined.

6.1.3 Testing of a microheater thermocycler

• Microheaters were designed and fabricated based on the power requirements

necessary to heat a reaction volume greater than 6 K/s.

• A Wheatstone bridge control system was designed and tested for the microheater.

• Heat treatment was done on the microheater to stabilize the resistance.

• The microheaters were tested to assess the performance under two different

conditions. The first condition was with fan cooling only with the heater

suspended in air. The second condition was with the heater mounted on top of an

aluminum heat sink and with fan cooling.

• 2-D heat transfer software was used to analyze the performance of thermally

bonded polycarbonate chambers on the 5 mm square microheater filled with

water.

• The pressure rise inside a sealed 5 mm square rigid chamber of depth 250 µm

filled 67 % with water and 33 % with air was modeled during heating from

ambient to 94 °C.

152

• Two different types of polycarbonate reaction vessels were tested to determine

their efficacy for the PCR reaction. One type of polycarbonate reaction vessel

consisted of three sheets of polycarbonate thermally bonded. The other type of

polycarbonate chamber was fabricated by drilling a hole into a 2 mm thick sheet

of polycarbonate. Some of these chambers were treated with an oxygen plasma

etch.

6.2 Conclusions

• PCR is influenced by the ability of the heater to heat the reaction vessel. It was

shown in Sections 2.2, 2.4, 3.2, and 4.5 that different heaters have different

heating and cooling rates. The average heating rate for the air in convective

devices or the surface of the heater for heat conduction devices varied from 1.2

K/s for the single fan convective thermocycler to 11.0 K/s for the microheater

with heat sink. The average cooling rate varied from 1.2 K/s for the convective

thermocycler to 8.0 K/s for the microheater mounted on a heat sink with fan

cooling. The air in the single fan convective thermocycler takes 33 seconds to

cool from 94 to 54 °C. The surface of the microheater can cool from 94 to 54 °C

in 5 seconds, based on the average cooling rate stated above. The single fan

convective thermocycler would spend 16 minutes cooling the air to the setpoint

over the course of a 30 cycle PCR reaction while the microheater would only

spend 2 minutes cooling to the setpoint over the course of a 30 cycle PCR

reaction. This demonstrates the effect that heating and cooling rates have on the

time to complete a PCR reaction.

153

• The results of the modeling of a constant wall thickness polypropylene capillary

in the convective thermocycler filled with water shows that the time constant

increases from 5 seconds for an 890 µm diameter capillary to 65 seconds for a

cylinder diameter of 5 mm with the same wall thickness and air velocity of 1.2

m/s as shown in Section 3.3. 2.12 mm outside diameter spherical reaction vessels

made of glass containing an equivalent volume of water and an equivalent wall

thickness to a cylindrical reaction vessel of outside diameter 890 µm and inside

diameter 400 µm were shown to have time constants 4 times larger then the

cylindrical case because of the larger surface area to volume ratio of the cylinder

as shown in Section 3.3. The analysis showed that reaction volumes of different

geometries for the convective thermocycler have different time constants. Time

constants for reaction chambers can be minimized by using capillaries with

outside diameters on the order of 900 µm. The advantage of using a convective

thermocycler is that different geometry reaction vessels may easily be placed into

them.

• Analysis done in Section 5.1 with 2-D finite element analysis showed that the

time constant for two nodes, one in the corner and one at the midline, in contact

with the reaction fluid at the top of a 5 mm square thermally bonded

polycarbonate chamber fabricated from three 250 µm thick sheets centered on the

5 mm square microheater were 4 seconds and 2.5 seconds, respectively. The time

constant for two nodes, one in the corner and one at the midline, in contact with

the reaction fluid at the top of a 3 mm square thermally bonded polycarbonate

chamber fabricated from three 250 µm thick sheets centered on the 5 mm square

154

microheater were 2.5 seconds and 2 seconds, respectively. Based on these results

it was concluded that square microchambers should have perimeters that lie inside

of the microheater for greater temperature uniformity.

• The time constant for two nodes, one in the corner and one at the midline, in

contact with the reaction fluid at the top of a 3 mm square thermally bonded

silicon chamber fabricated from three 250 µm thick sections centered on the 5

mm square microheater were 0.03 seconds and 0.01 seconds, respectively, as

analyzed in Section 5.1 with a 2-D heat transfer program. Based on this work, it

was concluded that changing the material of the reaction vessel to a material with

a higher thermal diffusivity decreased the time constant and provided for uniform

heating.

• Based on the PCR time study in Section 2.4.3 it was concluded that the time

limiting step in PCR is the polymerase nucleotide insertion rate. An 879 bp

segment of DNA could not be copied with a 25 second elongation time and a

denaturation and annealing time of 15 seconds. The 879 bp segment could be

copied with a 30 second insertion rate, however. This led to the conclusion that

the nucleotide insertion rate was 29 bp/s.

• The power consumption of an independent thermoelectric cooler used for PCR

can be decreased by 30 % not using a cooling fan on the heat sink with a loss in

average cooling of 1.5 K/s, based on the analysis done in Section 2.2. This would

amount to 1 minute of time for a 30 cycle PCR reaction. Based on this analysis,

the reduction in power consumption warrants not using the fan on the heat sink.

155

• Air heating rates for the double fan convective thermocycler built in this lab were

comparable to the heating rate of the surface of the TEC and consumed 80 times

more power on average based on the analysis done in Section 3.2. Convective

thermocyclers of the type built in this lab are not capable of low power operation

with high heating and cooling rates.

• The microheaters analyzed in Section 4.3.3 fabricated with liftoff and plastic

masks suffered delamination. Platinum elements with 30 µm feature size

deposited with a thickness of 1000 Å with a 150 Å chrome adhesion layer were

pitted, especially at the edges, while 50 µm features were more uniform. The

effective resistivity for the 30 µm elements was 3.6 x 10-7 Ω-m while the effective

resistivity for the 50 µm elements was 2.7 x 10-7 Ω-m. The bulk resistivity of

platinum is 1.0 x 10-7 Ω-m. Edge deformities and pitting influence the resistivity

of the deposited platinum.

• A 13 hour heat treatment at 250 °C reduced the resistivity of the temperature

probe from 2.7 x 10-7 Ω-m to 2.1 x 10-7 Ω-m based on the analysis in Section 4.4.

Heat treatment of platinum heaters is necessary to stabilize the resistance.

• The average heating rate for the microheater was 4 times faster then any of the

other thermal cycling devices tested as analyzed in Section 4.5. The average

cooling rate was 1.5 times faster than any of the other devices tested. The average

power for the microheater was 12 % of the power required by its nearest

competitor. Microheaters are low power, fast cycling devices.

• The cooling rate of a microheater can be increased by mounting it on a heat sink

with no change in average heating performance and double the average cooling

156

rate, based on the analysis in section 4.5.1. The average power required is

doubled from 0.5 W to 1.0 W, however. A heat sink should be used to speed

cooling of a microheater.

• The pressure rise during heating from ambient to 94 °C in a rigid, closed 5 mm

square thermally bonded reaction volume of height 250 µm filled with water and

an air bubble occupying 33% of the reaction volume approaches 100 kPa using an

ideal gas mixture model discussed in Section 5.1. Sealing of the reaction vessel is

possible with cyanoacrylate and polycarbonate.

• The pressure rise in a square reaction vessel with filling capillaries on the

microheater pushes the water out of the main reaction chamber and out into the

capillaries, discussed in Section 5.3.3. If the filling capillaries are completely full

then some of the reaction will remain in the central reaction volume if the ends of

the capillaries are securely sealed. The filling capillaries for the 5 mm square

thermally bonded polycarbonate chamber comprised of a middle section of

thickness 250 µm had a volume of 6 µL, equal to the central reaction chamber

volume. This large volume provides an area for fluid to flow out into during

heating unless they are completely full.

• Air bubble formation could be seen in a thermally bonded polycarbonate chamber

filled with ethylene glycol on the microheater at 94 °C with the capillaries sealed

with cyanoacrylate and polycarbonate, based on the observations described in

Section 5.3.3. Air bubbles present in chambers filled with water expanded. Since

ethylene glycol has a boiling point of 197 °C negligible ethylene glycol was

evaporating. The other source of air in the chamber is air trapped between the

157

liquid and the chamber surface, air trapped in the polycarbonate, and air within

the fluid.

• PCR was routinely completed in glass capillaries with only mineral oil for

sealing, discussed in section 5.3.2. One difference between glass and

polycarbonate is glass is hydrophilic. Hydrophilic glass capillaries can be sealed

with mineral oil and thermal cycled.

• Chambers drilled into polycarbonate without any surface treatment filled with

water and sealed with mineral oil did not perform well in cycling tests, while

oxygen plasma treated polycarbonate wells filled with water and sealed with

mineral oil were able to be thermal cycled without the solution evaporating based

on observations discussed in Section 5.3.1. The failure mechanism for the

untreated chambers was air bubbles rising to the surface and breaching the oil

layer. The shape of the oil-water interface in the well was significantly impacted

by the surface treatment. The treated well had a relatively uniform thickness of

oil over a largely flat water surface. Oxygen plasma treated polycarbonate

chambers sealed with mineral oil can be used for thermal cycling without

complete evaporation of the sample.

6.3 Future work

The microheaters show promise for rapid thermal cycling. The microheaters

fabricated suffered from delamination problems, especially smaller features. This

problem may have been caused by contamination of the e-beam reaction chamber during

158

evaporation. The resistance of the heaters continued to change even after heat treatment

at 250 °C for 13 hours. A longer, hotter heat treatment may be necessary.

The thermally bonded polycarbonate chambers did not perform well in holding the

solution in the central reaction chamber. Polycarbonate is an attractive choice for making

reaction vessels because it can be bonded in many different ways including tape, glue,

and thermally. It can be cut and drilled very easily. Polycarbonate chambers with valves

at the inlet and outlet to the chamber could be used to contain the reaction mixture.

Another option is to make the volume of the capillaries leading up to the reaction vessel

very small so there is no room for the reaction mixture to expand into. Very small filling

holes should be utilized to minimize the pressure on the sealing method.

A detection method also needs to be implemented. It was shown in this thesis that

it is possible to run PCR with detection reagents in the reaction volume. A detection

scheme would involve an illumination source and a fluorescence detector.

A more complete system could be designed that consists of a chip that could be

inserted into an array of microheaters. This would involve designing a structure to

encase the microheaters and a system to position the reaction chambers on the

microchambers and hold them there securely. The control system also needs to be

miniaturized. A computer data acquisition unit is currently being used to collect data and

provide control. It is possible to implement this control scheme with a microchip. The

design needs to move away from plug in power supplies and toward portable battery

power.

159

REFERENCES Accuratus. (2005). "Accuratus aluminum oxide properties 96 % aluminum oxide."

Retrieved March 6, 2005, from http://www.accuratus.com/alumox.html. Alciatore, D. G. and M. B. Histand (2003). Introduction to Mechatronics and

Measurement Systems, McGraw-Hill Higher Education. Aryee, E., R. Bailey, et al. (2005). "Detection, quantification and genotyping of Herpes

Simplex Virus in cervicovaginal secretions by real-time PCR: a cross sectional survey." Virology Journal 2(61).

Benammar, M. and W. C. Maskell (1989). "Temperature control of thick-film printed

heaters." Journal of Physical Engineering: Scientific Instrumentation 22: 933-936. Bilitewski, U., M. Genrich, et al. (2003). "Biochemical analysis with microfluidic

systems." Analytical and Bionanalytical Chemistry 377: 556-569. Cadoret, J.-C., B. Rousseau, et al. (2005). "Cyclic Nucleotides, the Photosynthetic

Apparatus and Response to a UV-B Stress in the Cyanobacterium Synechocystis sp. PCC 6803." Journal of Biological Chemistry.

Chen, J., W. Musundi, et al. (2005). "Electrokinetically Synchronized Polymerase Chain

Reaction Microchip Fabricated in Polycarbonate." Analytical Chemistry 77: 658-666.

D'Amico, M. L., V. Paiotta, et al. (2002). "A Kinetic Study of the Intercalation of

Ethidium Bromide into Poly (A) Poly (U)." Journal of Physical Chemistry B 106: 12635-12641.

Daniel, J. H., S. Iqbal, et al. (1998). "Silicon microchambers for DNA amplification."

Sensors and Actuators A 71: 81-88. Draghici, S., P. Khatri, et al. (2005). Identification of genomic signatures for the design

of assays for the detection and monitoring of anthrax threats. Pacific Symposium on Biocomputing 2005, World Scientific.

Efunda. (2005). "Efunda polymer material properties." Retrieved July 8, 2005, from

http://www.efunda.com/materials/polymers/properties/polymer_datasheet.cfm?MajorID=PC&MinorID=11.

El-Ali, J., I. R. Perch-Nielsen, et al. (2004). "Simulation and Experimental Validation of

a SU-8 Based PCR Thermocycler Chip with Integrated Heaters and Temperature Sensor." Sensors and Actuators A 110: 3-10.

160

Fermer, C., P. Nilsson, et al. (2003). "Microwave-assisted high-speed PCR." European Journal of Pharmaceutical Sciences 18: 129-132.

Giani, A., F. Mailly, et al. (2002). "Investigationof Pt/Ti bilayer on SiNx/Si substrates for

thermal sensor applications." Journal of Vacuum Science Technology A 20(1): 112-116

. Gill, P. (2005). "DNA as Evidence-The Technology of Identification." The New England

Journal of Medicine 352(26): 2669-2671. Giordano, B. C., J. Ferrance, et al. (2001). "Polymerase Chain Reaction in Polymeric

Microchips: DNA Amplification in Less than 240 Seconds." Analytical Biochemistry 291: 124-132.

Goodfellow. (2005). "Goodfellow polypropylene material information." Retrieved June

16, 2005, from http://www.goodfellow.com/csp/active/STATIC/E/Polypropylene.HTML.

Gulliksen, A., L. A. Solli, et al. (2005). "Parallel nanoliter detection of cancer markers

using polymer microchips." Lab on a Chip 5: 416-420. Halverson, R. M., M. W. MacIntyre, et al. (1982). "The Mechanism of Single-Step

Liftoff with Chlorobenzene in a Diazo-Type Resist." IBM Journal of Research and Development 26(5): 590-595.

Hindson, B. J., A. J. Makarewicz, et al. (2005). "APDS: the autonomous pathogen

detection system." Biosensors and Bioelectronics 20: 1925-1931. Incropera, F. and D. DeWitt (1996). Fundamentals of Heat and Mass Transfer, John

Wiley & Sons. Kermekchiev, M. B., A. Tzekov, et al. (2003). "Cold-sensitive mutants of Taq DNA

polymerase provide a hot start for PCR." Nucleic Acids Research 31(21): 6139-6147.

Klintberg, L., M. Svedberg, et al. (2003). "Fabrication of a paraffin actuator using hot

embossing of polycarbonate." Sensors and Actuators A 103(3): 307-316. Kopp, M. U., A. J. de Mello, et al. (1998). "Chemical Amplification: Continuous-Flow

PCR on a Chip." Science 280: 1046-1047. Kovacs, G. (1998). Micromachined Transducers Sourcebook, McGraw-Hill. Lane, S., J. Everman, et al. (2004). "Amplicon secondary structure prevents target

hybridization to oligonucleotide microarrays." Biosensors and Bioelectronics 20(4): 728-735.

161

Larsson, A. and H. Derand (2001). "Stability of Polycarbonate and Polystyrene Surfaces

after Hydrophilization with High Intensity Oxygen RF Plasma." Journal of Colloid and Interface Science 246: 214-221.

Lee, D. S., S. Ho Park, et al. (2004). "Bulk-micromachined submicroliter-volume PCR

chip with very rapid thermal response and low power consumption." Lab on a Chip 4: 401-407.

Lee, D. S., C. Y. Tsai, et al. (2004). "A new thermal cycling mechanism for effective

polymerase chain reaction in microliter volumes." Microsystem Technologies 10: 579-584.

LightCycler. (2005). "Technical Aspects of the LightCycler Instrument." Retrieved July

14, 2005, from http://www.roche-applied-science.com/sis/rtpcr/lightcycler/. Liu, Y., D. Ganser, et al. (2001). "Microfabricated Polycarbonate CE Devices for DNA

Analysis." Analytical Chemistry 73(17): 4196-4201. Liu, Y., C. Rauch, et al. (2002). "DNA Amplification and Hybridization Assays in

Integrated Plastic Monolithic Devices." Analytical Chemistry 74: 3063-6070. Loctite. (2000). "Loctite Laboratory Data Sheet, Product 190848." Retrieved August 9,

2005. Loeffler, J., N. Henke, et al. (2000). "Quantification of Fungal DNA by Using

Fluorescence Resonance Energy Transfer and the Light Cycler System." Journal of Clinical Microbiology 38(2): 586-590.

Lourenco, M. J., J. M. Serra, et al. (1998). "Thin Film Characterization for High

Temperature Applications." International Journal of Thermophysics 19(4): 1253-1265.

Martin, P. M., D. W. Matson, et al. (1998). Fabrication of plastic microfluidic

components. Conference on Microfluidics Devices and Systems, Santa Clara, CA, The International Society for Optical Engineering.

Matsubara, Y., K. Kerman, et al. (2004). "On-Chip Nanoliter-Volume Multiplex TaqMan

Polymerase Chain Reaction from a Single Copy Based on Counting Fluorescence Released Microchambers." Analytical Chemistry 76(21): 6434-6439.

Matsubara, Y., K. Kerman, et al. (2005). "Microchamber array based DNA quantification

and specific sequence detection from a single copy via PCR in nanoliter volumes." Biosensors and Bioelectronics 20: 1482-1490.

162

Melcor. (2005). "Melcor thermal solutions." Retrieved July 23, 2005, from http://www.melcor.com/pdf/MTScat.pdf.

Meldrum, D., H. Evensen, et al. (2000). "Acapella-1K, A Capillary-Based Submicroliter

Automated Fluid Handling System for Genome Analysis." Genome Research 10(1): 95-104.

Moran, M. J. and H. N. Shapiro (1999). Fundamentals of Engineering Thermodynamics,

John Wiley & Sons, Inc. Nagai, H., Y. Murakami, et al. (2001). "Development of a microchamber array for

picoliter PCR." Analytical Chemistry 73(5): 1043-1047 Nagai, H., Y. Murakami, et al. (2000). "High-throughput PCR in silicon based

microchamber array." Biosensors & Bioelectronics 16: 1015-1019. Nam, H.-M., V. Srinivasan, et al. (2005). "Application of SYBR green real-time PCR

assay for specific detection of Salmonella spp. in dairy farm environmental samples." International Journal of Food Microbiology 102: 161-171.

Nitsche, A., N. Steur, et al. (1999). "Different Real-Time PCR Formats Compared for the

Quantitative Detection of Human Cytomegalovirus DNA." Clinical Chemistry 45(11): 1932-1937.

Pyun, C.-H. and S.-M. Park (1985). Fluorescence Enhancement Mechanisms of Ethidium

Bromide by DNA or DNA Base Molecules. The Electrochemical Society Extended Abstracts Fall Meeting, Las Vegas, NV, Electrochemical Society.

Rotron, C. (2005). "Comair Rotron Patriot AC fan." Retrieved July 5, 2005, from

http://www.comairrotron.com/images/curves/patriotac3350.pdf. Rottmayer, R. E. and R. W. Hoffman (1971). "Structure and Intrinsic Stress of Platinum

Films." Journal of Vacuum Science Technology A 8(1): 151-155. Scheinert, P., B. Behrens, et al. (2005). "Optimizing DNA Amplification Protocols Using

the Eppendorf Mastercycler." Retrieved July 6, 2005, from http://www.eppendorfna.com/applications/PCR_appl_protocolsMC.asp.

Selvaganapathy, P. R., E. T. Carlen, et al. (2003). "Recent Progress in Microfluidic

Devices for Nucleic Acid and Antibody Assays." Proceedings of the IEEE 91(6): 954-975.

Shoffner, M. A., J. Cheng, et al. (1996). "Chip PCR. I. Surface passivation of

microfabricated silicon-glass chips for PCR." Nucleic Acids Research 24(2): 375-379.

163

Vierstraete, A. (2004). "PCR: Polymerase Chain Reaction." Retrieved June 1, 2005, from http://users.ugent.be/~avierstr/principles/pcr.html.

Winter, M. (2005). "Web Elements Periodic Table." Retrieved July 14, 2005, from

http://www.webelements.com. Wittwer, C. T., G. C. Fillmore, et al. (1989). "Automated polymerase chain reaction in

capillary tubes with hot air." Nucleic Acids Research 17(11): 4353-4357. Yang, J., Y. Liu, et al. (2002). "High sensitivity PCR assay in plastic micro reactors." Lab

on a Chip 2: 179-187. Ye, M.-Y., X.-F. Yin, et al. (2005). "DNA separation with low-viscosity sieving matrix

on microfabricated polycarbonate microfluidic chips." Analytical and Bionanalytical Chemistry 381: 820-827.

Yoon, D. S., Y.-S. Lee, et al. (2002). "Precise temperature control and rapid thermal

cycling in a micromachined DNA polymerase chain reaction." Journal of Micromechanics and Microengineering 12(6): 813-823.

Zhao, Z., D. Cui, et al. (2002). "An integrated biochip design and fabrication."

Proceedings of SPIE-The International Society for Optical Engineering 4936: 321-326.

Zhao, Z., Z. Cui, et al. (2003). "Monolithically integrated PCR biochip for DNA

amplification." Sensors and Actuators A 108: 162-167. Zou, Q., Y. Miao, et al. (2002). "Micro-Assembled Multi-Chamber Thermal Cycler for

Low-Cost Reaction Chip Thermal Multiplexing." Sensors and Actuators A 102: 114-121.

ABSTRACT Title of Thesis: THERMAL CYCLING DESIGN ...

Documents

Transcript of ABSTRACT Title of Thesis: THERMAL CYCLING DESIGN ...